defer few imports

4 files changed, 15 insertions, 5 deletions

diff --git a/src/sec_certs/dataset/cc.py b/src/sec_certs/dataset/cc.py
index 910a0a6a..c2d9ca0c 100644
--- a/src/sec_certs/dataset/cc.py
+++ b/src/sec_certs/dataset/cc.py
@@ -12,7 +12,6 @@ from typing import ClassVar, Iterator, cast
 import numpy as np
 import pandas as pd
 import requests
-import tabula
 from bs4 import BeautifulSoup, NavigableString, Tag
 
 import sec_certs.utils.sanitization
@@ -1532,6 +1531,8 @@ class CCSchemeDataset:
 
     @staticmethod
     def get_turkey_certified():
+        import tabula
+
         results = []
         with tempfile.TemporaryDirectory() as tmpdir:
             pdf_path = Path(tmpdir) / "turkey.pdf"
diff --git a/src/sec_certs/model/cpe_matching.py b/src/sec_certs/model/cpe_matching.py
index 5d08d7af..3015aeaa 100644
--- a/src/sec_certs/model/cpe_matching.py
+++ b/src/sec_certs/model/cpe_matching.py
@@ -6,7 +6,6 @@ import operator
 import re
 from typing import Pattern
 
-import spacy
 from rapidfuzz import fuzz
 
 from sec_certs import cert_rules, constants
@@ -28,6 +27,8 @@ class CPEClassifier:
     vendors_: set[str]
 
     def __init__(self, match_threshold: int = 80, n_max_matches: int = 10, spacy_model_to_use: str = "en_core_web_sm"):
+        import spacy
+
         self.match_threshold = match_threshold
         self.n_max_matches = n_max_matches
         self.nlp = spacy.load(spacy_model_to_use, disable=["parser", "ner"])
diff --git a/src/sec_certs/sample/fips.py b/src/sec_certs/sample/fips.py
index 82aa1c18..d351c3ce 100644
--- a/src/sec_certs/sample/fips.py
+++ b/src/sec_certs/sample/fips.py
@@ -12,7 +12,6 @@ import numpy as np
 import pandas as pd
 import requests
 from bs4 import BeautifulSoup, Tag
-from tabula import read_pdf
 
 from sec_certs import constants
 from sec_certs.cert_rules import FIPS_ALGS_IN_TABLE, fips_rules
@@ -632,6 +631,8 @@ class FIPSCertificate(
         Retrieves IDs of algorithms from tables inside security policy pdfs.
         External library is used to handle this.
         """
+        from tabula import read_pdf
+
         if table_rich_page_numbers := tables.find_pages_with_tables(cert.state.policy_txt_path):
             pdf.repair_pdf(cert.state.policy_pdf_path)
             try:
diff --git a/src/sec_certs/utils/pdf.py b/src/sec_certs/utils/pdf.py
index edda0570..749a8a5a 100644
--- a/src/sec_certs/utils/pdf.py
+++ b/src/sec_certs/utils/pdf.py
@@ -11,8 +11,6 @@ from typing import Any
 
 import pdftotext
 import pikepdf
-from pypdf import PdfReader
-from pypdf.generic import BooleanObject, ByteStringObject, FloatObject, IndirectObject, NumberObject, TextStringObject
 
 from sec_certs import constants
 from sec_certs.constants import (
@@ -156,6 +154,15 @@ def extract_pdf_metadata(filepath: Path) -> tuple[str, dict[str, Any] | None]:
     :param filepath: THe path to the PDF.
     :return: A tuple of the result code (see constants) and the metadata dictionary.
     """
+    from pypdf import PdfReader
+    from pypdf.generic import (
+        BooleanObject,
+        ByteStringObject,
+        FloatObject,
+        IndirectObject,
+        NumberObject,
+        TextStringObject,
+    )
 
     def map_metadata_value(val, nope_out=False):
         if isinstance(val, BooleanObject):