Reduce dependencies for license scanning API

src/licensedcode/query.py

@@ -17,7 +17,6 @@
 
 from intbitset import intbitset
 
-import typecode
 
 from commoncode.text import toascii
 from licensedcode.spans import Span
@@ -120,6 +119,7 @@ def build_query(
     Return a Query built from location or query string given an index.
     """
     if location:
+        import typecode
         T = typecode.get_type(location)
         # TODO: implement additional type-driven heuristics for query chunking.
         if not T.contains_text:
@@ -584,6 +584,7 @@ def _tokenize_and_build_runs(self, tokens_by_line, line_threshold=4):
         query_runs_append = self.query_runs.append
 
         if self.location:
+            import typecode
             ft = typecode.get_type(self.location)
             if ft.is_text_with_long_lines:
                 self.has_long_lines = True

src/scancode/api.py

@@ -12,10 +12,10 @@
 import os
 import sys
 
+from license_expression import combine_expressions
 from commoncode.filetype import get_last_modified_date
 from commoncode.hash import multi_checksums
 from scancode import ScancodeError
-from typecode.contenttype import get_type
 
 TRACE = os.environ.get('SCANCODE_DEBUG_API', False)
 
@@ -179,7 +179,6 @@ def get_licenses(
     from licensedcode.cache import build_spdx_license_expression
     from licensedcode.cache import get_cache
     from licensedcode.detection import detect_licenses
-    from packagedcode.utils import combine_expressions
 
     license_clues = []
     license_detections = []
@@ -340,6 +339,7 @@ def get_file_info(location, **kwargs):
     """
     Return a mapping of file information collected for the file at `location`.
     """
+    from typecode.contenttype import get_type
     result = {}
 
     # TODO: move date and size these to the inventory collection step???

src/textcode/analysis.py

@@ -14,7 +14,6 @@
 import unicodedata
 
 import chardet
-import typecode
 
 from textcode import pdf
 from textcode import markup
@@ -86,6 +85,7 @@ def numbered_text_lines(
             logger_debug('numbered_text_lines:', 'plain_text')
         return enumerate(unicode_text_lines(location), start_line)
 
+    import typecode
     T = typecode.get_type(location)
 
     if TRACE:
@@ -184,6 +184,7 @@ def unicode_text_lines_from_binary(location):
     Return an iterable over unicode text lines extracted from a binary file at
     location.
     """
+    import typecode
     T = typecode.get_type(location)
     if T.contains_text:
         for line in strings.strings_from_file(location):

tests/scancode/test_api.py

@@ -8,6 +8,8 @@
 #
 
 import os
+import subprocess
+import sys
 
 from commoncode.testcase import FileBasedTesting
 from scancode import api
@@ -144,3 +146,13 @@ def test_get_license_returns_correct_lines(self):
         assert results['detected_license_expression'] == 'mit'
         assert results['license_detections'][0]['matches'][0]['start_line'] == 2
         assert results['license_detections'][0]['matches'][0]['end_line'] == 4
+
+    def test_get_licenses_imports_without_heavy_dependencies(self):
+        script = (
+            'import sys; '
+            'from scancode.api import get_licenses; '
+            'assert "typecode" not in sys.modules; '
+            'assert "extractcode" not in sys.modules; '
+            'assert "packagedcode" not in sys.modules'
+        )
+        subprocess.check_call([sys.executable, '-c', script])