drewcassidy/yaclog
1
0
Fork 0
mirror of https://github.com/drewcassidy/yaclog.git synced 2024-09-01 14:58:58 +00:00
Code Issues Projects Releases Wiki Activity

reworked 2-step parser that can handle setext headers

Browse Source
This commit is contained in:
Andrew Cassidy 2021-04-18 18:45:39 -07:00
parent fa97e9154b
commit 849438a5f5
1 changed files with 106 additions and 53 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

159
yaclog/changelog.py
View File

@ -23,6 +23,13 @@ from typing import List, Tuple, Optional
bullets = '+-*' bullets = '+-*'
brackets = '[]' brackets = '[]'
code_regex = re.compile(r'^```')
header_regex = re.compile(r'^(?P<hashes>#+)\s+(?P<contents>[^#]+)(?:\s+#+)?$')
under1_regex = re.compile(r'^=+\s*$')
under2_regex = re.compile(r'^-+\s*$')
bullet_regex = re.compile(r'^[-+*]')
linkid_regex = re.compile(r'^\[(?P<link_id>\S*)]:\s*(?P<link>.*)')
def _strip_link(token): def _strip_link(token):
if link_literal := re.fullmatch(r'\[(.*?)]\((.*?)\)', token): if link_literal := re.fullmatch(r'\[(.*?)]\((.*?)\)', token):
@ -36,6 +43,22 @@ def _strip_link(token):
return token, None, None return token, None, None
def _join_markdown(segments: List[str]) -> str:
text: List[str] = []
last_bullet = False
for segment in segments:
is_bullet = bullet_regex.match(segment) and '\n' not in segment
if not is_bullet or not last_bullet:
text.append('')
text.append(segment)
last_bullet = is_bullet
return '\n'.join(text).strip()
class VersionEntry: class VersionEntry:
def __init__(self): def __init__(self):
self.sections = {'': []} self.sections = {'': []}
@ -64,7 +87,6 @@ class Changelog:
self.path = path self.path = path
self.header = '' self.header = ''
self.versions = [] self.versions = []
self.links = {}
# Read file # Read file
with open(path, 'r') as fp: with open(path, 'r') as fp:
@ -74,69 +96,104 @@ class Changelog:
in_block = False in_block = False
in_code = False in_code = False
# loop over lines in the file self.links = {}
links = {}
segments: List[Tuple[int, List[str], str]] = []
header_segments = []
for line_no, line in enumerate(self.lines): for line_no, line in enumerate(self.lines):
if in_code: if in_code:
if re.match(r'^```', line): # this is the contents of a code block
line = '```' segments[-1][1].append(line)
if code_regex.match(line):
in_code = False in_code = False
in_block = False in_block = False
if len(self.versions) == 0: elif code_regex.match(line):
self.header += line # this is the start of a code block
else:
self.versions[-1].sections[section][-1] += line
elif re.match(r'^```', line):
in_code = True in_code = True
if len(self.versions) == 0: segments.append((line_no, [line], 'code'))
self.header += line
else:
self.versions[-1].sections[section].append(line)
elif match := re.fullmatch( elif under1_regex.match(line) and in_block and len(segments[-1][1]) == 1 and segments[-1][2] == 'p':
r'^##\s+(?P<name>\S*)(?:\s+-\s+(?P<date>\S+))?\s*?(?P<extra>.*?)\s*#*$', line): # this is an underline for a setext-style H1
# this is a version header in the form '## Name (- date) (tags*) (#*)' # ugly but it works
section = '' last = segments.pop()
segments.append((last[0], last[1] + [line], 'h1'))
elif under2_regex.match(line) and in_block and len(segments[-1][1]) == 1 and segments[-1][2] == 'p':
# this is an underline for a setext-style H2
# ugly but it works
last = segments.pop()
segments.append((last[0], last[1] + [line], 'h2'))
elif bullet_regex.match(line):
in_block = True
segments.append((line_no, [line], 'li'))
elif match := header_regex.match(line):
# this is a header
kind = f'h{len(match["hashes"])}'
segments.append((line_no, [line], kind))
in_block = False in_block = False
self._add_version_header(match['name'], match['date'], match['extra'], line_no)
elif match := re.fullmatch(r'\[(\S*)]:\s*(\S*)\n', line): elif match := linkid_regex.match(line):
# this is a link definition in the form '[id]: link', so add it to the link table # this is a link definition in the form '[id]: link', so add it to the link table
self.links[match[1].lower()] = match[2] links[match['link_id'].lower()] = match['link']
elif len(self.versions) == 0:
# we haven't encountered any version headers yet,
# so its best to just add this line to the header string
self.header += line
elif line.isspace(): elif line.isspace():
# skip empty lines # skip empty lines
in_block = False in_block = False
pass
elif match := re.fullmatch(r'###\s+(\S*?)(\s+#*)?', line):
# this is a version section header in the form '### Name' or '### Name ###'
section = match[1].title()
if section not in self.versions[-1].sections.keys():
self.versions[-1].sections[section] = []
in_block = False
elif line[0] in '+-*#':
# bullet point or subheader
# subheaders are mostly preserved for round-trip accuracy, and are discouraged in favor of bullet points
# bullet points are preserved since some people like to use '+', '-' or '*' for different things
self.versions[-1].sections[section].append(line.strip())
in_block = True
elif in_block: elif in_block:
# not a bullet point, header, etc, and in a block, so this line should be appended to the last # this is a line to be added to a paragraph
self.versions[-1].sections[section][-1] += '\n' + line.strip() segments[-1][1].append(line)
else:
# this is a new paragraph
in_block = True
segments.append((line_no, [line], 'p'))
for segment in segments:
text = ''.join(segment[1]).strip()
if segment[2] == 'h2':
# start of a version
split = text.rstrip('-').strip('#').strip().split()
if '-' in split:
split.remove('-')
version = VersionEntry()
section = ''
version.name, version.link, version.link_id = _strip_link(split[0])
version.line_no = segment[0]
if len(split) > 1:
try:
version.date = datetime.date.fromisoformat(split[1].strip(string.punctuation))
except ValueError:
version.date = None
if len(split) > 2:
version.tags = [s.strip('[]') for s in split[2:]]
self.versions.append(version)
elif len(self.versions) == 0:
# we haven't encountered any version headers yet,
# so its best to just add this line to the header string
header_segments.append(text)
elif segment[2] == 'h3':
# start of a version section
section = text.strip('#').strip()
if section not in self.versions[-1].sections.keys():
self.versions[-1].sections[section] = []
else: else:
# not a bullet point, header, etc, and not in a block, so this is the start of a new paragraph # change log entry
self.versions[-1].sections[section].append(line.strip()) self.versions[-1].sections[section].append(text)
in_block = True
# handle links # handle links
for version in self.versions: for version in self.versions:
@ -153,7 +210,7 @@ class Changelog:
version.link = self.links.pop(version.link_id) version.link = self.links.pop(version.link_id)
# strip whitespace from header # strip whitespace from header
self.header = self.header.strip() self.header = _join_markdown(header_segments)
def write(self, path: os.PathLike = None): def write(self, path: os.PathLike = None):
if path is None: if path is None:
@ -176,13 +233,9 @@ class Changelog:
if section: if section:
fp.write(f'### {section}\n\n') fp.write(f'### {section}\n\n')
for entry in version.sections[section]:
fp.write(entry + '\n')
if entry[0] not in '-+*':
fp.write('\n')
if len(version.sections[section]) > 0: if len(version.sections[section]) > 0:
fp.write('\n') fp.write(_join_markdown(version.sections[section]))
fp.write('\n\n')
for link_id, link in v_links.items(): for link_id, link in v_links.items():
fp.write(f'[{link_id.lower()}]: {link}\n') fp.write(f'[{link_id.lower()}]: {link}\n')