Детализирована документация. Небольшие изменения имен параметров.

index1c/parse_modules.py

@@ -0,0 +1,89 @@
+"""Индексация BSL-модулей конфигурации."""
+
+from __future__ import annotations
+
+import re
+from dataclasses import dataclass
+from pathlib import Path
+
+from .config import OBJECT_TYPE_FOLDERS
+
+# Процедура/Функция Имя(...)
+RE_SYMBOL = re.compile(
+    r"(?im)^\s*(?:&[А-Яа-яA-Za-z_]+)*\s*"
+    r"(?:Процедура|Функция|Procedure|Function)\s+"
+    r"([А-Яа-яA-Za-z_][А-Яа-яA-Za-z0-9_]*)"
+)
+
+# Лимит тела для FTS (символы) — полные гигантские модули не кладём целиком.
+MAX_BODY_CHARS = 200_000
+
+
+@dataclass
+class ModuleRec:
+    config: str
+    rel_path: str
+    abs_path: str
+    owner: str
+    module_kind: str
+    symbols: list[str]
+    body: str
+    line_count: int
+    size: int
+    mtime_ns: int
+
+
+def _module_kind(rel: Path) -> str:
+    name = rel.name
+    if name == "Module.bsl":
+        parent = rel.parent.name
+        if parent == "Form":
+            return "FormModule"
+        if parent == "Ext":
+            # CommonModules/.../Ext/Module.bsl или HTTPServices/.../Ext/Module.bsl
+            return "Module"
+        return "Module"
+    if name.endswith("Module.bsl"):
+        return name[: -len(".bsl")]  # ObjectModule, ManagerModule, ...
+    return name
+
+
+def _owner_from_rel(rel: Path) -> str:
+    """Documents/ЗаказКлиента/Ext/ObjectModule.bsl → Document.ЗаказКлиента."""
+    parts = rel.parts
+    if not parts:
+        return ""
+    folder = parts[0]
+    otype = OBJECT_TYPE_FOLDERS.get(folder)
+    if otype and len(parts) >= 2:
+        return f"{otype}.{parts[1]}"
+    return folder
+
+
+def parse_bsl_file(path: Path, *, config: str, src_root: Path) -> ModuleRec | None:
+    try:
+        st = path.stat()
+        text = path.read_text(encoding="utf-8-sig", errors="replace")
+    except OSError:
+        return None
+    rel = path.relative_to(src_root).as_posix()
+    symbols = sorted(set(RE_SYMBOL.findall(text)))
+    body = text if len(text) <= MAX_BODY_CHARS else text[:MAX_BODY_CHARS]
+    return ModuleRec(
+        config=config,
+        rel_path=rel,
+        abs_path=str(path),
+        owner=_owner_from_rel(Path(rel)),
+        module_kind=_module_kind(Path(rel)),
+        symbols=symbols,
+        body=body,
+        line_count=text.count("\n") + (1 if text and not text.endswith("\n") else 0),
+        size=st.st_size,
+        mtime_ns=getattr(st, "st_mtime_ns", int(st.st_mtime * 1e9)),
+    )
+
+
+def iter_bsl_files(src: Path) -> list[Path]:
+    if not src.is_dir():
+        return []
+    return sorted(src.rglob("*.bsl"))