3857244254

Build Type

github

Committed by GitHub

Commit Message

Merge 204223951 into a035bf0e5

Pull Request Pull Request #341: Feature/UI for language download

Run Details

49 of 163 new or added lines in 9 files covered. (30.06%)

22 existing lines in 2 files now uncovered.

1711 of 2183 relevant lines covered (78.38%)

2.23 hits per line

Source File
Press 'n' to go to next uncovered line, 'b' for previous

87.1

/src/normcap/ocr/utils.py

import functools
import logging
import os
import traceback
from typing import Optional

from pytesseract import pytesseract

from normcap.version import Version

logger = logging.getLogger(__name__)


def tsv_to_list_of_dicts(tsv_data: dict) -> list[dict]:
    """Transpose tsv dict from k:list[v] to list[Dict[k:v]]."""
    words: list[dict] = [{} for _ in tsv_data["level"]]
    for k, values in tsv_data.items():
        for idx, v in enumerate(values):
            words[idx][k] = v

    # Filter empty words
    return [w for w in words if w["text"].strip()]


def get_tesseract_config(tessdata_path: Optional[os.PathLike]) -> str:
    """Get string with cli args to be passed into tesseract api."""
    return f'--tessdata-dir "{tessdata_path}"' if tessdata_path else ""


def get_tesseract_languages(
    tesseract_cmd: os.PathLike, tessdata_path: Optional[os.PathLike]
) -> list[str]:
    """Get info abput tesseract setup."""
    pytesseract.tesseract_cmd = str(tesseract_cmd)

    try:
        languages = sorted(
            pytesseract.get_languages(config=get_tesseract_config(tessdata_path))
        )
    except RuntimeError as e:
        traceback.print_tb(e.__traceback__)
        raise RuntimeError(
            "Couldn't determine Tesseract information. If you pip installed NormCap "
            + "make sure Tesseract is installed and configured correctly."
        ) from e

    if not languages:
        raise ValueError(
            "Could not load any languages for tesseract. "
            + "On Windows, make sure that TESSDATA_PREFIX environment variable is set. "
            + "On Linux/macOS see if 'tesseract --list-langs' work is the command line."
        )

    return languages


@functools.lru_cache
def get_tesseract_version(tesseract_cmd: os.PathLike) -> Version:
    """Get info abput tesseract setup."""
    pytesseract.tesseract_cmd = str(tesseract_cmd)
    tesseract_version = str(pytesseract.get_tesseract_version()).splitlines()[0]
    return Version(tesseract_version)

1	import functools	3✔
2	import logging	3✔
3	import os	3✔
4	import traceback	3✔
5	from typing import Optional	3✔
6
7	from pytesseract import pytesseract	3✔
8
9	from normcap.version import Version	3✔
10
11	logger = logging.getLogger(__name__)	3✔
12
13
14	def tsv_to_list_of_dicts(tsv_data: dict) -> list[dict]:	3✔
15	"""Transpose tsv dict from k:list[v] to list[Dict[k:v]]."""
16	words: list[dict] = [{} for _ in tsv_data["level"]]	3✔
17	for k, values in tsv_data.items():	3✔
18	for idx, v in enumerate(values):	3✔
19	words[idx][k] = v	3✔
20
21	# Filter empty words
22	return [w for w in words if w["text"].strip()]	3✔
23
24
25	def get_tesseract_config(tessdata_path: Optional[os.PathLike]) -> str:	3✔
26	"""Get string with cli args to be passed into tesseract api."""
27	return f'--tessdata-dir "{tessdata_path}"' if tessdata_path else ""	3✔
28
29
30	def get_tesseract_languages(	3✔
31	tesseract_cmd: os.PathLike, tessdata_path: Optional[os.PathLike]
32	) -> list[str]:
33	"""Get info abput tesseract setup."""
34	pytesseract.tesseract_cmd = str(tesseract_cmd)	3✔
35
36	try:	3✔
37	languages = sorted(	3✔
38	pytesseract.get_languages(config=get_tesseract_config(tessdata_path))
39	)
40	except RuntimeError as e:	×
41	traceback.print_tb(e.__traceback__)	×
42	raise RuntimeError(	×
43	"Couldn't determine Tesseract information. If you pip installed NormCap "
44	+ "make sure Tesseract is installed and configured correctly."
45	) from e
46
47	if not languages:	3✔
UNCOV 48	raise ValueError(	×
49	"Could not load any languages for tesseract. "
50	+ "On Windows, make sure that TESSDATA_PREFIX environment variable is set. "
51	+ "On Linux/macOS see if 'tesseract --list-langs' work is the command line."
52	)
53
54	return languages	3✔
55
56
57	@functools.lru_cache	3✔
58	def get_tesseract_version(tesseract_cmd: os.PathLike) -> Version:	3✔
59	"""Get info abput tesseract setup."""
60	pytesseract.tesseract_cmd = str(tesseract_cmd)	3✔
61	tesseract_version = str(pytesseract.get_tesseract_version()).splitlines()[0]	3✔
62	return Version(tesseract_version)	3✔

dynobo / normcap / 3857244254

Source File Press 'n' to go to next uncovered line, 'b' for previous

Source File
Press 'n' to go to next uncovered line, 'b' for previous