rework tool to work with the sqlite database
This commit is contained in:
parent
c66f0a5409
commit
49db7931a3
115
tlfi/__main__.py
115
tlfi/__main__.py
|
|
@ -7,12 +7,13 @@ formatting almost like in the TLFi.
|
|||
|
||||
import argparse
|
||||
import difflib
|
||||
import gzip
|
||||
import lzma
|
||||
import os
|
||||
import re
|
||||
import unicodedata
|
||||
from functools import cache
|
||||
from os import environ
|
||||
import sqlite3
|
||||
import sys
|
||||
from pathlib import Path
|
||||
from typing import Generator
|
||||
|
||||
from bs4 import BeautifulSoup, NavigableString
|
||||
|
||||
|
|
@ -25,80 +26,53 @@ except ImportError:
|
|||
return ""
|
||||
t = DummyTerminal()
|
||||
|
||||
TAG_STRIP_RE = re.compile(r"\s+")
|
||||
|
||||
|
||||
def main():
|
||||
ap = argparse.ArgumentParser(description="TLFi CLI")
|
||||
ap.add_argument("query", help="mot(s) à chercher")
|
||||
ap.add_argument("-f", "--lexical-forms",
|
||||
default=get_root_path() / "lexical_forms.txt",
|
||||
help="fichier des formes lexicales")
|
||||
ap.add_argument("-d", "--definitions",
|
||||
default=get_root_path() / "definitions",
|
||||
help="répertoire des définitions")
|
||||
ap.add_argument("-d", "--database", help="base de données")
|
||||
args = ap.parse_args()
|
||||
|
||||
lookup_res = lookup(args.query, args.lexical_forms)
|
||||
if lookup_res is None:
|
||||
exit()
|
||||
if args.database:
|
||||
db_path = Path(args.database)
|
||||
elif (env_db_path := os.environ.get("TLFI_DATABASE")):
|
||||
db_path = Path(env_db_path)
|
||||
else:
|
||||
sys.exit("Pas de base de données et TLFI_DATABASE n'est pas défini.")
|
||||
|
||||
if (defs := get_definition_paths(lookup_res, args.definitions)):
|
||||
for d in defs:
|
||||
show_definition(d)
|
||||
lexical_form = args.query
|
||||
connection = sqlite3.connect(db_path)
|
||||
printed = 0
|
||||
for definition_html in get_definitions(lexical_form, connection):
|
||||
pretty_print_definition(definition_html)
|
||||
printed += 1
|
||||
if printed == 0:
|
||||
print("Forme lexicale non trouvée. Suggestions :")
|
||||
for suggestion in get_suggestions(lexical_form, connection):
|
||||
print(f"* {suggestion}")
|
||||
|
||||
|
||||
@cache
|
||||
def get_root_path():
|
||||
return Path(environ.get("TLFI_ROOT", "."))
|
||||
|
||||
|
||||
def lookup(query, lexical_forms_path):
|
||||
"""Return a form for which a definition might exist, else None.
|
||||
|
||||
If we are sure the lexical form does not have definitions, suggest similar
|
||||
words to the user.
|
||||
"""
|
||||
with open(lexical_forms_path, "rt") as lexical_forms_file:
|
||||
forms = lexical_forms_file.readlines()
|
||||
if query + "\n" in forms:
|
||||
return query
|
||||
|
||||
print("Suggestions :")
|
||||
suggestions = (
|
||||
form for form in map(str.rstrip, forms)
|
||||
if difflib.SequenceMatcher(None, query, form).ratio() > 0.8
|
||||
def get_definitions(
|
||||
lexical_form: str,
|
||||
connection: sqlite3.Connection
|
||||
) -> Generator[str, None, None]:
|
||||
cursor = connection.cursor()
|
||||
result = cursor.execute(
|
||||
"SELECT definition FROM definitions WHERE lexical_form = ?",
|
||||
(lexical_form,)
|
||||
)
|
||||
for form in suggestions:
|
||||
print(f"- {form}")
|
||||
return None
|
||||
for (blob,) in result.fetchall():
|
||||
yield lzma.decompress(blob).decode()
|
||||
|
||||
|
||||
def get_definition_paths(query, definitions) -> list:
|
||||
"""Return a list of definition file paths for this lexical form."""
|
||||
nfkd = unicodedata.normalize("NFKD", query[0])
|
||||
first_char = next((c for c in nfkd if not unicodedata.combining(c)), "")
|
||||
if not first_char:
|
||||
print("Pas compris la première lettre…")
|
||||
return []
|
||||
path = Path(definitions) / first_char.upper() / query
|
||||
try:
|
||||
return [f for f in path.iterdir() if str(f).endswith(".txt.gz")]
|
||||
except FileNotFoundError:
|
||||
print("Définition non trouvée.")
|
||||
return []
|
||||
|
||||
|
||||
def show_definition(def_path):
|
||||
"""Print a definition from a definition file."""
|
||||
with gzip.open(def_path, "rt") as def_file:
|
||||
html = def_file.read()
|
||||
def pretty_print_definition(html: str):
|
||||
soup = BeautifulSoup(html, "html.parser")
|
||||
content = parse_tag(soup.div.div)
|
||||
print(content)
|
||||
|
||||
|
||||
TAG_STRIP_RE = re.compile(r"\s+")
|
||||
|
||||
|
||||
def parse_tag(tag) -> str:
|
||||
if isinstance(tag, NavigableString):
|
||||
return TAG_STRIP_RE.sub(" ", tag)
|
||||
|
|
@ -124,5 +98,24 @@ def parse_tag(tag) -> str:
|
|||
return content
|
||||
|
||||
|
||||
def get_suggestions(
|
||||
query: str,
|
||||
connection: sqlite3.Connection
|
||||
) -> Generator[str, None, None]:
|
||||
"""Return a form for which a definition might exist, else None.
|
||||
|
||||
If we are sure the lexical form does not have definitions, suggest similar
|
||||
words to the user.
|
||||
"""
|
||||
cursor = connection.cursor()
|
||||
result = cursor.execute("SELECT lexical_form FROM definitions")
|
||||
yielded = []
|
||||
for (form,) in result.fetchall():
|
||||
if difflib.SequenceMatcher(None, query, form).ratio() > 0.8:
|
||||
if form not in yielded:
|
||||
yield form
|
||||
yielded.append(form)
|
||||
|
||||
|
||||
if __name__ == "__main__":
|
||||
main()
|
||||
|
|
|
|||
Loading…
Reference in a new issue