Coverage for little_loops / frontmatter.py: 11%

1"""Frontmatter read/write utilities for little-loops. 

2 

3Provides shared YAML-subset frontmatter parsing, stripping, and updating 

4used by issue_parser, sync, and issue_history modules. 

5""" 

6 

7from __future__ import annotations 

8 

9import logging 

10import re 

11import textwrap 

12from typing import Any 

13 

14import yaml 

15 

16logger = logging.getLogger(__name__) 

17 

18STATUS_SYNONYMS: dict[str, str] = { 

19 "complete": "done", 

20 "completed": "done", 

21 "finished": "done", 

22 "closed": "done", 

23 "in-progress": "in_progress", 

24 "in progress": "in_progress", 

25 "wip": "in_progress", 

26} 

27 

28 

29def parse_frontmatter(content: str, *, coerce_types: bool = False) -> dict[str, Any]: 

30 """Extract YAML frontmatter from content. 

31 

32 Looks for content between opening and closing '---' markers. 

33 Parses a subset of YAML: simple ``key: value`` pairs, YAML block 

34 sequences (``key:`` followed by ``- item`` lines), and block scalars 

35 (``key: |`` or ``key: >`` followed by indented lines). Nested 

36 structures are not supported and will emit a ``logging.WARNING``. 

37 Returns empty dict if no frontmatter found. 

38 

39 Args: 

40 content: File content to parse 

41 coerce_types: If True, coerce digit strings to int 

42 

43 Returns: 

44 Dictionary of frontmatter fields, or empty dict 

45 """ 

46 if not content or not content.startswith("---"): 

47 return {} 

48 

49 end_match = re.search(r"\n---\s*\n", content[3:]) 

50 if not end_match: 

51 return {} 

52 

53 frontmatter_text = content[4 : 3 + end_match.start()] 

54 

55 result: dict[str, Any] = {} 

56 current_list_key: str | None = None 

57 lines = frontmatter_text.split("\n") 

58 i = 0 

59 while i < len(lines): 

60 line = lines[i].strip() 

61 i += 1 

62 if not line or line.startswith("#"): 

63 continue 

64 if line.startswith("- "): 

65 if current_list_key is not None: 

66 result[current_list_key].append(line[2:].strip()) 

67 else: 

68 logger.warning("Unsupported YAML list syntax in frontmatter: %r", line) 

69 continue 

70 # Non-list line: finalize any in-progress empty list, then reset 

71 if current_list_key is not None and result[current_list_key] == []: 

72 result[current_list_key] = None 

73 current_list_key = None 

74 if ":" in line: 

75 key, value = line.split(":", 1) 

76 key = key.strip() 

77 value = value.strip() 

78 if value.startswith("|") or value.startswith(">"): 

79 # Block scalar: collect indented continuation lines 

80 block_type = value[0] 

81 block_lines: list[str] = [] 

82 while i < len(lines): 

83 next_line = lines[i] 

84 if next_line and (next_line[0] == " " or next_line[0] == "\t"): 

85 block_lines.append(next_line) 

86 i += 1 

87 else: 

88 break 

89 if block_lines: 

90 dedented = textwrap.dedent("\n".join(block_lines)) 

91 if block_type == ">": 

92 dedented = re.sub(r"\s+", " ", dedented).strip() 

93 result[key] = dedented 

94 else: 

95 result[key] = "" 

96 continue 

97 if value.startswith("[") and value.endswith("]"): 

98 inner = value[1:-1].strip() 

99 result[key] = [item.strip() for item in inner.split(",")] if inner else [] 

100 continue 

101 if value.lower() in ("null", "~", ""): 

102 if value == "": 

103 result[key] = [] 

104 current_list_key = key 

105 else: 

106 result[key] = None 

107 elif coerce_types and value.isdigit(): 

108 result[key] = int(value) 

109 else: 

110 # Strip surrounding YAML string quotes (single or double) 

111 if len(value) >= 2 and value[0] == value[-1] and value[0] in ("'", '"'): 

112 value = value[1:-1] 

113 result[key] = value 

114 # Finalize any trailing empty list key 

115 if current_list_key is not None and result[current_list_key] == []: 

116 result[current_list_key] = None 

117 if "status" in result and isinstance(result["status"], str): 

118 result["status"] = STATUS_SYNONYMS.get(result["status"], result["status"]) 

119 return result 

120 

121 

122def parse_skill_frontmatter(text: str) -> dict[str, str]: 

123 """Extract flat key/value pairs from SKILL.md frontmatter. 

124 

125 Uses ``yaml.safe_load`` so YAML block scalars (e.g. ``description: |``) 

126 are resolved to their string content instead of the indicator literal. 

127 Non-string scalar values are stringified; nested structures are dropped. 

128 

129 If the frontmatter is not valid YAML (e.g. unquoted colons in values), 

130 falls back to a permissive line-based scan — top-level ``key: value`` 

131 pairs only, block scalars are not resolved in that path. 

132 

133 This is the canonical SKILL.md frontmatter parser. Prefer it over the 

134 general ``parse_frontmatter`` for SKILL.md files: ``parse_frontmatter`` 

135 deliberately drops block scalars (logs a warning, sets value to 

136 ``None``) which loses the description body for skills that use 

137 ``description: |``. 

138 """ 

139 if not text.startswith("---"): 

140 return {} 

141 end = text.find("---", 3) 

142 if end == -1: 

143 return {} 

144 fm_text = text[3:end] 

145 try: 

146 loaded = yaml.safe_load(fm_text) 

147 except yaml.YAMLError: 

148 loaded = None 

149 if isinstance(loaded, dict): 

150 fm: dict[str, str] = {} 

151 for key, value in loaded.items(): 

152 if value is None: 

153 fm[str(key)] = "" 

154 elif isinstance(value, str): 

155 fm[str(key)] = value 

156 elif isinstance(value, bool | int | float): 

157 fm[str(key)] = str(value).lower() if isinstance(value, bool) else str(value) 

158 return fm 

159 fm = {} 

160 for line in fm_text.splitlines(): 

161 if line and not line.startswith(" ") and not line.startswith("\t") and ":" in line: 

162 key, _, val = line.partition(":") 

163 fm[key.strip()] = val.strip() 

164 return fm 

165 

166 

167def strip_frontmatter(content: str) -> str: 

168 """Remove YAML frontmatter from content, returning the body. 

169 

170 Strips the ``---`` delimited frontmatter block (if present) and 

171 returns everything after the closing delimiter. 

172 

173 Args: 

174 content: File content possibly starting with frontmatter 

175 

176 Returns: 

177 Content with frontmatter removed. Returns original content 

178 unchanged if no valid frontmatter block is found. 

179 """ 

180 if not content or not content.startswith("---"): 

181 return content 

182 

183 end_match = re.search(r"\n---\s*\n", content[3:]) 

184 if not end_match: 

185 return content 

186 

187 return content[3 + end_match.end() :] 

188 

189 

190def update_frontmatter(content: str, updates: dict[str, Any]) -> str: 

191 """Update or add frontmatter fields in content. 

192 

193 Merges ``updates`` into an existing ``---`` delimited YAML frontmatter 

194 block, preserving other fields and their order. If no frontmatter block 

195 exists, a new one is prepended. Existing keys are overwritten with the 

196 new values. 

197 

198 Args: 

199 content: Full file content, possibly with existing frontmatter 

200 updates: Fields to add/update in frontmatter; values may be nested dicts 

201 

202 Returns: 

203 Content with updated frontmatter block 

204 """ 

205 fm_match = re.match(r"^---\n(.*?)\n---", content, re.DOTALL) 

206 if not fm_match: 

207 fm_text = yaml.dump(dict(updates), default_flow_style=False, sort_keys=False).strip() 

208 return f"---\n{fm_text}\n---\n{content}" 

209 

210 existing: dict[str, Any] = yaml.safe_load(fm_match.group(1)) or {} 

211 existing.update(updates) 

212 fm_text = yaml.dump(existing, default_flow_style=False, sort_keys=False).strip() 

213 return f"---\n{fm_text}\n---{content[fm_match.end() :]}"