Dates and exclusions

ampoule_ssg/__init__.py

@@ -1,16 +1,24 @@

                                
                                
                                
                            
                                
                                    
                                        
                                            1  
                                            #!/usr/bin/env python3
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            2  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            3  
                                            import os
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        4 -
                                        import jinja2
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        5 -
                                        from ruamel.yaml import YAML
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        4 +
                                        import re
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            5  
                                            import shutil
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            6  
                                            import contextlib
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        7 +
                                        from datetime import datetime
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        8 +
                                        import jinja2
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        9 +
                                        from ruamel.yaml import YAML
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            10  
                                            import colorama
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            11  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            12  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            13  
                                            colorama.init()
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            14  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            15  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        16 +
                                        # Disable YAML date constructor
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        17 +
                                        def no_date_constructor(loader, node):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        18 +
                                            value = loader.construct_scalar(node)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        19 +
                                            return value
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        20 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        21 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            22  
                                            @contextlib.contextmanager
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            23  
                                            def in_directory(directory):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            24  
                                                cwd = os.getcwd()
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -29,13 +37,95 @@ def delete_directory_contents(directory):

                                
                                
                                
                            
                                
                                    
                                        
                                            37  
                                                        shutil.rmtree(os.path.join(root, dir))
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            38  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            39  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        40 +
                                        def parse_date_string(date_string):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        41 +
                                            def split_date_and_time(date_string):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        42 +
                                                if ":" not in date_string:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        43 +
                                                    return date_string, "00:00:00"
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        44 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        45 +
                                                elements = date_string.partition(":")
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        46 +
                                                partition_character = " "
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        47 +
                                                if " " not in date_string:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        48 +
                                                    partition_character = "-"
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        49 +
                                                    if "-" not in date_string:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        50 +
                                                        partition_character = "T"
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        51 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        52 +
                                                date = elements[0].rpartition(partition_character)[0].strip()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        53 +
                                                time = elements[0].rpartition(partition_character)[2].strip() + elements[1] + elements[2].strip()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        54 +
                                                time = time.removeprefix("T").removesuffix("Z")
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        55 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        56 +
                                                return date, time
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        57 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        58 +
                                            time_formats = [
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        59 +
                                                # 24-hour ISO
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        60 +
                                                "%H:%M:%S",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        61 +
                                                "%H:%M",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        62 +
                                                "%H",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        63 +
                                                # Single digit hour
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        64 +
                                                "-%H:%M:%S",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        65 +
                                                "-%H:%M",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        66 +
                                                "-%H",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        67 +
                                                # 12-hour (AM/PM)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        68 +
                                                "%I:%M:%S %p",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        69 +
                                                "%I:%M %p",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        70 +
                                                "%I %p",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        71 +
                                                # Single digit 12-hour
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        72 +
                                                "-%I:%M:%S %p",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        73 +
                                                "-%I:%M %p",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        74 +
                                                "-%I %p",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        75 +
                                            ]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        76 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        77 +
                                            date_formats = [
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        78 +
                                                # ISO formats
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        79 +
                                                "%Y-%m-%d",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        80 +
                                                "%y-%m-%d",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        81 +
                                                # European formats
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        82 +
                                                "%d.%m.%Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        83 +
                                                "%d.%m.%y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        84 +
                                                # American formats
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        85 +
                                                "%m/%d/%Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        86 +
                                                "%m/%d/%y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        87 +
                                                # Text-based European formats
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        88 +
                                                "%d %B %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        89 +
                                                "%d %b %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        90 +
                                                "%d %B, %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        91 +
                                                "%d %b, %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        92 +
                                                # Text-based American formats
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        93 +
                                                "%B %d %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        94 +
                                                "%b %d %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        95 +
                                                "%B %d, %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        96 +
                                                "%b %d, %Y",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        97 +
                                                # ISO weekly calendar
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        98 +
                                                "%G-W%V-%u",
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        99 +
                                            ]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        100 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        101 +
                                            date, time = split_date_and_time(date_string)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        102 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        103 +
                                            time_object = datetime.min.time()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        104 +
                                            date_object = datetime.min.date()
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        105 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        106 +
                                            for time_format in time_formats:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        107 +
                                                try:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        108 +
                                                    time_object = datetime.strptime(time, time_format)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        109 +
                                                except ValueError:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        110 +
                                                    pass
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        111 +
                                            for date_format in date_formats:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        112 +
                                                try:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        113 +
                                                    date_object = datetime.strptime(date, date_format)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        114 +
                                                except ValueError:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        115 +
                                                    pass
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        116 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        117 +
                                            return datetime.combine(date_object, time_object.time())
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        118 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        119 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            120  
                                            class Document:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            121  
                                                def __init__(self, file_name, url_transform=lambda x: x):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            122  
                                                    self.file_name = file_name
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            123  
                                                    self.encoding = "utf-8"
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            124  
                                                    # If the file is text, read it.
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            125  
                                                    self.front_matter = YAML()
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        126 +
                                                self.front_matter.Constructor.add_constructor("tag:yaml.org,2002:timestamp", no_date_constructor)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            127  
                                                    self.content = ""
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        128 +
                                                self.date = datetime.fromtimestamp(os.path.getmtime(file_name))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            129  
                                                    try:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            130  
                                                        with open(file_name, "r", encoding=self.encoding) as f:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            131  
                                                            print(colorama.Style.RESET_ALL, colorama.Style.BRIGHT, colorama.Fore.LIGHTWHITE_EX, f"Loading document {file_name}".ljust(shutil.get_terminal_size().columns), sep="")
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -51,10 +141,14 @@ class Document:

                                
                                
                                
                            
                                
                                    
                                        
                                            141  
                                                                    if line != "---\n":
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            142  
                                                                        front_matter += line
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            143  
                                                                print(colorama.Style.RESET_ALL, colorama.Fore.GREEN, "Front matter loaded", sep="")
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        54 -
                                                            print(front_matter)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            144  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            145  
                                                            if front_matter:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            146  
                                                                self.front_matter = self.front_matter.load(front_matter)
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        147 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        148 +
                                                            print(self.front_matter, type(self.front_matter))
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        149 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        150 +
                                                            if "DATE" in self.front_matter:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        151 +
                                                                self.date = parse_date_string(self.front_matter["DATE"])
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            152  
                                                            else:   # put it back
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            153  
                                                                self.content = initial_line
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            154  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        

@@ -88,15 +182,20 @@ class Document:

                                
                                
                                
                            
                                
                                    
                                        
                                            182  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            183  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            184  
                                            class Index:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        91 -
                                            def __init__(self, directory, recursive=False, url_transform=lambda x: x):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        185 +
                                            def __init__(self, directory, recursive=False, url_transform=lambda x: x, sort_by=lambda x: x.file_name, exclude=None):
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            186  
                                                    self.directory = directory
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            187  
                                                    # Temporarily move to the specified directory in order to read the files.
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        188 +
                                                if exclude:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        189 +
                                                    regex = re.compile(exclude)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        190 +
                                                else:
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        191 +
                                                    regex = re.compile("(?!)")
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            192  
                                                    with in_directory(directory):
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            193  
                                                        if recursive:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        96 -
                                                        self.file_names = [os.path.join(dir_path, f) for dir_path, dir_name, filenames in os.walk(".") for f in filenames]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        194 +
                                                        self.file_names = [os.path.join(dir_path, f) for dir_path, dir_name, filenames in os.walk(".") for f in filenames if not regex.search(f)]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            195  
                                                        else:
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        98 -
                                                        self.file_names = [i for i in os.listdir() if os.path.isfile(i)]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        99 -
                                                    self.documents = [Document(i, url_transform) for i in self.file_names]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        196 +
                                                        self.file_names = [i for i in os.listdir() if os.path.isfile(i) and not regex.search(i)]
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        197 +
                                        
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        198 +
                                                    self.documents = sorted([Document(i, url_transform) for i in self.file_names], key=sort_by)
                                        
                                        
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            199  
                                                    self.__current_index = 0
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            200  
                                            
                                        
                                        
                                            
                                            
                                            
                                            
                                        
                                    
                                
                                
                                
                            
                                
                                    
                                        
                                            201  
                                                def __iter__(self):