From 83056a4b6fda29c6c1e6caab5dc60c52c4474e6a Mon Sep 17 00:00:00 2001
From: Alfredo Deza <adeza@anchore.com>
Date: Tue, 21 Jul 2020 14:25:51 -0400
Subject: [PATCH] cataloger: add a requirements.txt parser

Signed-off-by: Alfredo Deza <adeza@anchore.com>
---
 imgbom/cataloger/python/parse_requirements.go | 69 +++++++++++++++++++
 1 file changed, 69 insertions(+)
 create mode 100644 imgbom/cataloger/python/parse_requirements.go

diff --git a/imgbom/cataloger/python/parse_requirements.go b/imgbom/cataloger/python/parse_requirements.go
new file mode 100644
index 000000000..decf25d88
--- /dev/null
+++ b/imgbom/cataloger/python/parse_requirements.go
@@ -0,0 +1,69 @@
+package python
+
+import (
+	"bufio"
+	"fmt"
+	"io"
+	"strings"
+
+	"github.com/anchore/imgbom/imgbom/pkg"
+)
+
+func parseRequirementsTxt(_ string, reader io.Reader) ([]pkg.Package, error) {
+	packages := make([]pkg.Package, 0)
+
+	scanner := bufio.NewScanner(reader)
+	for scanner.Scan() {
+		line := scanner.Text()
+
+		line = strings.TrimRight(line, "\n")
+
+		switch {
+		case strings.HasPrefix(line, "#"):
+			// commented out line, skip
+			continue
+		case strings.HasPrefix(line, "-e"):
+			// editable packages aren't parsed (yet)
+			continue
+		case len(strings.Split(line, "==")) < 2:
+			// a package without a version, or a range (unpinned) which
+			// does not tell us exactly what will be installed
+			// XXX only needed if we want to log this, otherwise the next case catches it
+			continue
+		case len(strings.Split(line, "==")) == 2:
+			// remove comments if present
+			uncommented := removeTrailingComment(line)
+			// parse a new requirement
+			parts := strings.Split(uncommented, "==")
+			name := strings.TrimSpace(parts[0])
+			version := strings.TrimSpace(parts[1])
+			packages = append(packages, pkg.Package{
+				Name:     name,
+				Version:  version,
+				Language: pkg.Python,
+				Type:     pkg.PythonRequirementsPkg,
+			})
+		default:
+			continue
+		}
+	}
+
+	if err := scanner.Err(); err != nil {
+		return nil, fmt.Errorf("failed to parse python requirements file: %w", err)
+	}
+
+	return packages, nil
+}
+
+func removeTrailingComment(line string) string {
+	parts := strings.Split(line, "#")
+	switch len(parts) {
+	case 1:
+		// there aren't any comments
+		return line
+	default:
+		// any number of "#" means we only want the first part, assuming this
+		// isn't prefixed with "#" (up to the caller)
+		return parts[0]
+	}
+}