[ie/youtube] Fix comments extraction (#9775)

[yt-dlp.git] / devscripts / make_readme.py

diff --git a/devscripts/make_readme.py b/devscripts/make_readme.py
index 15c4a7c7dc481f590facb9e064733dfa270b8222..2270b31d3b97a35f00a537ea54077d5252d32831 100755 (executable)
--- a/devscripts/make_readme.py
+++ b/devscripts/make_readme.py
@@ -1,16 +1,30 @@
 #!/usr/bin/env python3
 
-# yt-dlp --help | make_readme.py
-# This must be run in a console of correct width
+"""
+yt-dlp --help | make_readme.py
+This must be run in a console of correct width
+"""
+
+# Allow direct execution
+import os
+import sys
+
+sys.path.insert(0, os.path.dirname(os.path.dirname(os.path.abspath(__file__))))
+
+
 import functools
 import re
-import sys
+
+from devscripts.utils import read_file, write_file
 
 README_FILE = 'README.md'
 
 OPTIONS_START = 'General Options:'
 OPTIONS_END = 'CONFIGURATION'
 EPILOG_START = 'See full documentation'
+ALLOWED_OVERSHOOT = 2
+
+DISABLE_PATCH = object()
 
 
 def take_section(text, start=None, end=None, *, shift=0):
@@ -21,36 +35,59 @@ def take_section(text, start=None, end=None, *, shift=0):
 
 
 def apply_patch(text, patch):
-    return re.sub(*patch, text)
+    return text if patch[0] is DISABLE_PATCH else re.sub(*patch, text)
 
 
 options = take_section(sys.stdin.read(), f'\n  {OPTIONS_START}', f'\n{EPILOG_START}', shift=1)
 
+max_width = max(map(len, options.split('\n')))
 switch_col_width = len(re.search(r'(?m)^\s{5,}', options).group())
 delim = f'\n{" " * switch_col_width}'
 
 PATCHES = (
-    (  # Headings
+    (   # Standardize `--update` message
+        r'(?m)^(    -U, --update\s+).+(\n    \s.+)*$',
+        r'\1Update this program to the latest version',
+    ),
+    (   # Headings
         r'(?m)^  (\w.+\n)(    (?=\w))?',
         r'## \1'
     ),
-    (  # Do not split URLs
+    (   # Fixup `--date` formatting
+        rf'(?m)(    --date DATE.+({delim}[^\[]+)*)\[.+({delim}.+)*$',
+        (rf'\1[now|today|yesterday][-N[day|week|month|year]].{delim}'
+         f'E.g. "--date today-2weeks" downloads only{delim}'
+         'videos uploaded on the same day two weeks ago'),
+    ),
+    (   # Do not split URLs
         rf'({delim[:-1]})? (?P<label>\[\S+\] )?(?P<url>https?({delim})?:({delim})?/({delim})?/(({delim})?\S+)+)\s',
         lambda mobj: ''.join((delim, mobj.group('label') or '', re.sub(r'\s+', '', mobj.group('url')), '\n'))
     ),
-    # This creates issues with prepare_manpage
-    # (  # Avoid newline when a space is available b/w switch and description
-    #     r'(?m)^(\s{4}-.{%d})(%s)' % (switch_col_width - 6, delim),
-    #     r'\1 '
-    # ),
+    (   # Do not split "words"
+        rf'(?m)({delim}\S+)+$',
+        lambda mobj: ''.join((delim, mobj.group(0).replace(delim, '')))
+    ),
+    (   # Allow overshooting last line
+        rf'(?m)^(?P<prev>.+)${delim}(?P<current>.+)$(?!{delim})',
+        lambda mobj: (mobj.group().replace(delim, ' ')
+                      if len(mobj.group()) - len(delim) + 1 <= max_width + ALLOWED_OVERSHOOT
+                      else mobj.group())
+    ),
+    (   # Avoid newline when a space is available b/w switch and description
+        DISABLE_PATCH,  # This creates issues with prepare_manpage
+        r'(?m)^(\s{4}-.{%d})(%s)' % (switch_col_width - 6, delim),
+        r'\1 '
+    ),
+    (   # Replace brackets with a Markdown link
+        r'SponsorBlock API \((http.+)\)',
+        r'[SponsorBlock API](\1)'
+    ),
 )
 
-with open(README_FILE, encoding='utf-8') as f:
-    readme = f.read()
+readme = read_file(README_FILE)
 
-with open(README_FILE, 'w', encoding='utf-8') as f:
-    f.write(''.join((
-        take_section(readme, end=f'## {OPTIONS_START}'),
-        functools.reduce(apply_patch, PATCHES, options),
-        take_section(readme, f'# {OPTIONS_END}'),
-    )))
+write_file(README_FILE, ''.join((
+    take_section(readme, end=f'## {OPTIONS_START}'),
+    functools.reduce(apply_patch, PATCHES, options),
+    take_section(readme, f'# {OPTIONS_END}'),
+)))