Book a Demo!
CoCalc Logo Icon
StoreFeaturesDocsShareSupportNewsAboutPoliciesSign UpSign In
probml
GitHub Repository: probml/pyprobml
 Path: blob/master/internal/book1/figure_no_notebook_mapping.py
1192 views
1
"""

2
command usage:

3
python3 internal/book1/figure_no_notebook_mapping.py -lof internal/book1.lof -csv internal/book1/figures_nb_mapping_book1.csv

4
"""

5
import argparse

6
from probml_utils.url_utils import (

7
    dict_to_csv,

8
    extract_scripts_name_from_caption,

9
)

10
from TexSoup import TexSoup

11


12
parser = argparse.ArgumentParser(description="create figure url mapping")

13
parser.add_argument("-lof", "--lof", type=str, help="", default="internal/book2.lof")

14
parser.add_argument("-csv", "--csv", type=str, help="", default="internal/figures_url_mapping_book2.csv")

15
args = parser.parse_args()

16


17
BOOK_NO = int(args.lof.split("/")[-1].split(".")[0][-1])  # from internal/book1.lof to 1

18
NOTEBOOKS_MD_URL = "https://probml.github.io/notebooks#"

19


20
def fig_no_nb_mapping(lof_file_path, csv_name, make_url = False):

21
    f"""

22
    create mapping of fig_no to url by parsing lof_file and save mapping in {csv_name}

23
    """

24
    with open(lof_file_path) as fp:

25
        LoF_File_Contents = fp.read()

26
    soup = TexSoup(LoF_File_Contents)

27


28
    # create mapping of fig_no to list of script_name

29


30
    url_mapping = {}

31
    for caption in soup.find_all("numberline"):

32
        fig_no = str(caption.contents[0])

33
        extracted_scripts = extract_scripts_name_from_caption(str(caption))

34
        nb = None

35
        if len(extracted_scripts) == 1:

36
            nb = extracted_scripts[0]

37


38
        elif len(extracted_scripts) > 1: # use dummy notebooks

39
            chap, fig = fig_no.split(".")

40
            nb = f"fig_{chap}_{fig}.ipynb"

41
        

42
        if nb:

43
            if make_url: 

44
                url_mapping[fig_no] = NOTEBOOKS_MD_URL + nb

45
            else:

46
                url_mapping[fig_no] = nb

47


48


49
    if csv_name:

50
        dict_to_csv(url_mapping, csv_name)

51
    print(f"Mapping of {len(url_mapping)} urls is saved in {csv_name}")

52
    return url_mapping

53


54


55
print(f"Parsing started from {args.lof}...........")

56


57
if args.csv:

58
    figure_mapping = fig_no_nb_mapping(args.lof, args.csv)

59
    figure_mapping = fig_no_nb_mapping(args.lof, args.csv.replace(".csv","")+"_urls.csv", make_url=True)

60


61
else:

62
    figure_mapping = fig_no_nb_mapping(args.lof, "")

63
    figure_mapping = fig_no_nb_mapping(args.lof, args.csv.replace("csv","")+"_urls.csv", make_url=True)

64


65


66