fix: setup.py update using script

MANIFEST.in

@@ -5,3 +5,4 @@ include LICENSE.txt
 include README.rst
 include requirements/base.in
 recursive-include code_annotations *.html *.png *.gif *js *.css *jpg *jpeg *svg *py *.yaml *.yml
+include requirements/constraints.txt

setup.py

@@ -27,33 +27,95 @@ def load_requirements(*requirements_paths):
     """
     Load all requirements from the specified requirements files.
 
-    Returns:
-        list: Requirements file relative path strings
+    Requirements will include any constraints from files specified
+    with -c in the requirements files.
+    Returns a list of requirement strings.
     """
-    requirements = set()
+    # UPDATED VIA SEMGREP - if you need to remove/modify this method remove this line and add a comment specifying why.
+
+    # e.g. {"django": "Django", "confluent-kafka": "confluent_kafka[avro]"}
+    by_canonical_name = {}
+
+    def check_name_consistent(package):
+        """
+        Raise exception if package is named different ways.
+
+        This ensures that packages are named consistently so we can match
+        constraints to packages. It also ensures that if we require a package
+        with extras we don't constrain it without mentioning the extras (since
+        that too would interfere with matching constraints.)
+        """
+        canonical = package.lower().replace('_', '-').split('[')[0]
+        seen_spelling = by_canonical_name.get(canonical)
+        if seen_spelling is None:
+            by_canonical_name[canonical] = package
+        elif seen_spelling != package:
+            raise Exception(
+                f'Encountered both "{seen_spelling}" and "{package}" in requirements '
+                'and constraints files; please use just one or the other.'
+            )
+
+    requirements = {}
+    constraint_files = set()
+
+    # groups "pkg<=x.y.z,..." into ("pkg", "<=x.y.z,...")
+    re_package_name_base_chars = r"a-zA-Z0-9\-_."  # chars allowed in base package name
+    # Two groups: name[maybe,extras], and optionally a constraint
+    requirement_line_regex = re.compile(
+        r"([%s]+(?:\[[%s,\s]+\])?)([<>=][^#\s]+)?"
+        % (re_package_name_base_chars, re_package_name_base_chars)
+    )
+
+    def add_version_constraint_or_raise(current_line, current_requirements, add_if_not_present):
+        regex_match = requirement_line_regex.match(current_line)
+        if regex_match:
+            package = regex_match.group(1)
+            version_constraints = regex_match.group(2)
+            check_name_consistent(package)
+            existing_version_constraints = current_requirements.get(package, None)
+            # It's fine to add constraints to an unconstrained package,
+            # but raise an error if there are already constraints in place.
+            if existing_version_constraints and existing_version_constraints != version_constraints:
+                raise BaseException(f'Multiple constraint definitions found for {package}:'
+                                    f' "{existing_version_constraints}" and "{version_constraints}".'
+                                    f'Combine constraints into one location with {package}'
+                                    f'{existing_version_constraints},{version_constraints}.')
+            if add_if_not_present or package in current_requirements:
+                current_requirements[package] = version_constraints
+
+    # Read requirements from .in files and store the path to any
+    # constraint files that are pulled in.
     for path in requirements_paths:
-        requirements.update(
-            line.split('#')[0].strip() for line in open(path).readlines()
-            if is_requirement(line.strip())
-        )
-    return list(requirements)
+        with open(path) as reqs:
+            for line in reqs:
+                if is_requirement(line):
+                    add_version_constraint_or_raise(line, requirements, True)
+                if line and line.startswith('-c') and not line.startswith('-c http'):
+                    constraint_files.add(os.path.dirname(path) + '/' + line.split('#')[0].replace('-c', '').strip())
+
+    # process constraint files: add constraints to existing requirements
+    for constraint_file in constraint_files:
+        with open(constraint_file) as reader:
+            for line in reader:
+                if is_requirement(line):
+                    add_version_constraint_or_raise(line, requirements, False)
+
+    # process back into list of pkg><=constraints strings
+    constrained_requirements = [f'{pkg}{version or ""}' for (pkg, version) in sorted(requirements.items())]
+    return constrained_requirements
 
 
 def is_requirement(line):
     """
     Return True if the requirement line is a package requirement.
 
     Returns:
-        bool: True if the line is not blank, a comment, a URL, or an included file
+        bool: True if the line is not blank, a comment,
+        a URL, or an included file
     """
-    return not (
-        line == '' or
-        line.startswith('-r') or
-        line.startswith('#') or
-        line.startswith('-e') or
-        line.startswith('git+') or
-        line.startswith('-c')
-    )
+    # UPDATED VIA SEMGREP - if you need to remove/modify this method remove this line and add a comment specifying why
+
+    return line and line.strip() and not line.startswith(('-r', '#', '-e', 'git+', '-c'))
 
 
 VERSION = get_version('code_annotations', '__init__.py')