Daniel/usa
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

fix: 优化docker 镜像

Browse Source
This commit is contained in:
Daniel
2026-03-02 14:10:43 +08:00
parent 783a69dad1
commit 36576592a2
25 changed files with 491 additions and 58 deletions
Download Patch File Download Diff File Expand all files Collapse all files

54
crawler/extractor_rules.py
View File

@@ -24,18 +24,64 @@ def extract_from_news(text: str, timestamp: Optional[str] = None) -> Dict[str, A
t = (text or "").lower()
loss_us, loss_ir = {}, {}
v = _first_int(t, r"(?:us|american|u\.?s\.?)[\s\w]*(?:say|report)[\s\w]*(\d+)[\s\w]*(?:troop|soldier|killed|dead)")
# 美军人员伤亡
v = _first_int(t, r"(?:us|american|u\.?s\.?)[\s\w]*(?:say|report)[\s\w]*(\d+)[\s\w]*(?:troop|soldier|military)[\s\w]*(?:killed|dead)")
if v is not None:
loss_us["personnel_killed"] = v
v = _first_int(t, r"(\d+)[\s\w]*(?:us|american)[\s\w]*(?:troop|soldier|killed|dead)")
v = _first_int(t, r"(\d+)[\s\w]*(?:us|american)[\s\w]*(?:troop|soldier|military)[\s\w]*(?:killed|dead)")
if v is not None:
loss_us["personnel_killed"] = v
v = _first_int(t, r"(?:iran|iranian)[\s\w]*(?:say|report)[\s\w]*(\d+)[\s\w]*(?:troop|soldier|killed|dead)")
v = _first_int(t, r"(?:us|american)[\s\w]*(\d+)[\s\w]*(?:wounded|injured)")
if v is not None:
loss_us["personnel_wounded"] = v
# 伊朗人员伤亡
v = _first_int(t, r"(?:iran|iranian)[\s\w]*(?:say|report)[\s\w]*(\d+)[\s\w]*(?:troop|soldier|guard|killed|dead)")
if v is not None:
loss_ir["personnel_killed"] = v
v = _first_int(t, r"(\d+)[\s\w]*(?:iranian|iran)[\s\w]*(?:troop|soldier|killed|dead)")
v = _first_int(t, r"(\d+)[\s\w]*(?:iranian|iran)[\s\w]*(?:troop|soldier|guard|killed|dead)")
if v is not None:
loss_ir["personnel_killed"] = v
v = _first_int(t, r"(?:iran|iranian)[\s\w]*(\d+)[\s\w]*(?:wounded|injured)")
if v is not None:
loss_ir["personnel_wounded"] = v
# 平民伤亡（多不区分阵营，计入双方或仅 us 因多为美国基地周边）
v = _first_int(t, r"(\d+)[\s\w]*(?:civilian|civil)[\s\w]*(?:killed|dead)")
if v is not None:
loss_us["civilian_killed"] = v
v = _first_int(t, r"(\d+)[\s\w]*(?:civilian|civil)[\s\w]*(?:wounded|injured)")
if v is not None:
loss_us["civilian_wounded"] = v
# 基地损毁（美方基地居多）
v = _first_int(t, r"(\d+)[\s\w]*(?:base)[\s\w]*(?:destroyed|leveled)")
if v is not None:
loss_us["bases_destroyed"] = v
v = _first_int(t, r"(\d+)[\s\w]*(?:base)[\s\w]*(?:damaged|hit|struck)")
if v is not None:
loss_us["bases_damaged"] = v
if "base" in t and ("destroy" in t or "level" in t) and not loss_us.get("bases_destroyed"):
loss_us["bases_destroyed"] = 1
if "base" in t and ("damage" in t or "hit" in t or "struck" in t or "strike" in t) and not loss_us.get("bases_damaged"):
loss_us["bases_damaged"] = 1
# 战机 / 舰船（根据上下文判断阵营）
v = _first_int(t, r"(\d+)[\s\w]*(?:aircraft|plane|jet|fighter|f-?16|f-?35|f-?18)[\s\w]*(?:down|destroyed|lost|shot)")
if v is not None:
if "us" in t or "american" in t or "u.s" in t:
loss_us["aircraft"] = v
elif "iran" in t:
loss_ir["aircraft"] = v
else:
loss_us["aircraft"] = v
v = _first_int(t, r"(\d+)[\s\w]*(?:ship|destroyer|warship|vessel)[\s\w]*(?:hit|damaged|sunk)")
if v is not None:
if "iran" in t:
loss_ir["warships"] = v
else:
loss_us["warships"] = v
if loss_us:
out.setdefault("combat_losses_delta", {})["us"] = loss_us