Valsts-Pirmkods/Strategy-as-Code
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

NAP2027 0.1.1: labotas uzdevumu [315]–[318] ailes; pašreizējā institūcija (KEM)

Browse Source 
Uzdevumu teksta pirmie vārdi 68. lpp. bija nonākuši ailē „Nr.”; pārbaudei
pievienota apgrieztā vārdu pārbaude. Atribūts currentOrg/currentSince:
septiņiem VARAM vides un klimata uzdevumiem — KEM 20-0000.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_016679RwmHsuTFfxt26wP6rk
This commit is contained in:
Rihards Gailums
2026-10-11 13:01:14 +00:00
parent bcdd26b8d9
commit bc915bc6d2
10 changed files with 114 additions and 37 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
tools/parse_nap.py
View File

@@ -37,9 +37,9 @@ def load_vocab(pdf_path):
def _better(whole, *parts):
"""Join when the whole word is attested more often than any of its fragments."""
"""Join when the whole word is attested more often than any of its fragments, or once while a fragment never is."""
n = VOCAB.get(whole, 0) + VOCAB.get(whole.lower(), 0)
return n >= 2 and any(VOCAB.get(p, 0) < n for p in parts)
return (n >= 2 and any(VOCAB.get(p, 0) < n for p in parts)) or (n >= 1 and any(VOCAB.get(p, 0) == 0 for p in parts))
def repair(s):
@@ -50,7 +50,7 @@ def repair(s):
if out:
a, core = out[-1], t.rstrip(".,;:)”")
ca = a.lstrip("“(")
if "/" in a or "http" in a or "www." in a:
if "http" in a or "www." in a or ("/" in a and re.search(r"[_.%=?&]|/.*/|-$", a)):
out.append(t)
continue
# three fragments "fundamentāl a s"
@@ -101,7 +101,7 @@ def page_lines(pdf):
# footnotes: small text at the bottom of the page
marks = [w["top"] for w in ws if w["size"] < 6.5 and w["text"].isdigit() and w["x0"] < 90 and w["top"] > 550]
fz = min(marks) - 1.5 if marks else 9999
infoot = lambda w: w["size"] < 9.5 and w["top"] >= fz
infoot = lambda w: w["size"] < 10.5 and w["top"] >= fz
foot = [w for w in ws if infoot(w)]
body = [w for w in ws if not infoot(w) and w["size"] >= 7.5]
smalls = [w for w in ws if not infoot(w) and w["size"] < 7.5]
@@ -242,6 +242,7 @@ def assign(words, centers, names, bounds=None):
for w in words:
cx = (w["x0"] + w["x1"]) / 2
i = sum(1 for b in bounds if cx > b)
i = max(i, 1) # the row marker is not among the words: nothing belongs to the "Nr." column
cells[names[i]].append(w)
return cells