|
| 1 | +#!/usr/bin/env python3 |
| 2 | +"""Collapse duplicate top-level pins in pip-compile / pybuild-deps output. |
| 3 | +
|
| 4 | +``pybuild-deps compile`` can emit multiple ``name==version`` blocks for the same |
| 5 | +distribution when different build paths resolve different versions. pip then |
| 6 | +sees conflicting requirements; we keep the highest version and merge comments. |
| 7 | +""" |
| 8 | + |
| 9 | +from __future__ import annotations |
| 10 | + |
| 11 | +import sys |
| 12 | +from collections import defaultdict |
| 13 | +from pathlib import Path |
| 14 | + |
| 15 | +from packaging.utils import canonicalize_name |
| 16 | +from packaging.version import Version |
| 17 | + |
| 18 | + |
| 19 | +def parse_pin_line(line: str) -> tuple[str, str] | None: |
| 20 | + """Parse a single ``name==version`` line.""" |
| 21 | + s = line.strip() |
| 22 | + if not s or s.startswith("#"): |
| 23 | + return None |
| 24 | + if "==" not in s: |
| 25 | + return None |
| 26 | + name, _, ver = s.partition("==") |
| 27 | + name, ver = name.strip(), ver.strip() |
| 28 | + if not name or not ver: |
| 29 | + return None |
| 30 | + return name, ver |
| 31 | + |
| 32 | + |
| 33 | +def is_pin_line(line: str) -> bool: |
| 34 | + """Return True if ``line`` is a top-level pin (not a comment continuation).""" |
| 35 | + if not line.strip(): |
| 36 | + return False |
| 37 | + if line[0].isspace(): |
| 38 | + return False |
| 39 | + return parse_pin_line(line) is not None |
| 40 | + |
| 41 | + |
| 42 | +def split_segments(lines: list[str]) -> list[tuple[str, list[str]]]: |
| 43 | + """Split into ``('raw', ...)`` or ``('pin', block)`` segments.""" |
| 44 | + out: list[tuple[str, list[str]]] = [] |
| 45 | + i = 0 |
| 46 | + while i < len(lines): |
| 47 | + line = lines[i] |
| 48 | + if is_pin_line(line): |
| 49 | + block = [line] |
| 50 | + i += 1 |
| 51 | + while i < len(lines) and lines[i].startswith((" ", "\t")): |
| 52 | + block.append(lines[i]) |
| 53 | + i += 1 |
| 54 | + out.append(("pin", block)) |
| 55 | + else: |
| 56 | + out.append(("raw", [line])) |
| 57 | + i += 1 |
| 58 | + return out |
| 59 | + |
| 60 | + |
| 61 | +def dedupe_pin_blocks(blocks: list[list[str]]) -> list[str]: |
| 62 | + """Pick the highest version and merge ``# via`` comment lines from all blocks.""" |
| 63 | + |
| 64 | + def version_key(b: list[str]) -> Version: |
| 65 | + parsed = parse_pin_line(b[0]) |
| 66 | + assert parsed is not None |
| 67 | + return Version(parsed[1]) |
| 68 | + |
| 69 | + winner = max(blocks, key=version_key) |
| 70 | + merged_tail: list[str] = [] |
| 71 | + seen: set[str] = set() |
| 72 | + for b in blocks: |
| 73 | + for line in b[1:]: |
| 74 | + key = line.strip() |
| 75 | + if key and key not in seen: |
| 76 | + seen.add(key) |
| 77 | + merged_tail.append(line) |
| 78 | + return [winner[0], *merged_tail] |
| 79 | + |
| 80 | + |
| 81 | +def dedupe_file(path: Path) -> None: |
| 82 | + """Rewrite *path* in place with duplicate pins collapsed.""" |
| 83 | + lines = path.read_text().splitlines(keepends=True) |
| 84 | + segments = split_segments(lines) |
| 85 | + by_name: dict[str, list[list[str]]] = defaultdict(list) |
| 86 | + for kind, payload in segments: |
| 87 | + if kind != "pin": |
| 88 | + continue |
| 89 | + parsed = parse_pin_line(payload[0]) |
| 90 | + assert parsed is not None |
| 91 | + name, _ = parsed |
| 92 | + by_name[canonicalize_name(name)].append(payload) |
| 93 | + |
| 94 | + deduped: dict[str, list[str]] = { |
| 95 | + k: dedupe_pin_blocks(v) for k, v in by_name.items() |
| 96 | + } |
| 97 | + |
| 98 | + out: list[str] = [] |
| 99 | + seen_pin: set[str] = set() |
| 100 | + for kind, payload in segments: |
| 101 | + if kind == "raw": |
| 102 | + out.extend(payload) |
| 103 | + continue |
| 104 | + parsed = parse_pin_line(payload[0]) |
| 105 | + assert parsed is not None |
| 106 | + name, _ = parsed |
| 107 | + canon = canonicalize_name(name) |
| 108 | + if canon in seen_pin: |
| 109 | + continue |
| 110 | + seen_pin.add(canon) |
| 111 | + out.extend(deduped[canon]) |
| 112 | + |
| 113 | + path.write_text("".join(out)) |
| 114 | + |
| 115 | + |
| 116 | +def main() -> None: |
| 117 | + """CLI: single argument, path to ``requirements-build.txt``.""" |
| 118 | + if len(sys.argv) != 2: |
| 119 | + print( |
| 120 | + "usage: dedupe_requirements_build.py <requirements-build.txt>", |
| 121 | + file=sys.stderr, |
| 122 | + ) |
| 123 | + sys.exit(2) |
| 124 | + dedupe_file(Path(sys.argv[1])) |
| 125 | + |
| 126 | + |
| 127 | +if __name__ == "__main__": |
| 128 | + main() |
0 commit comments