Merge bucket-lines-{prob,hocr} into one tool called bucket-lines, that uses the filename extension to determine how to process the lines

3 files changed, 29 insertions, 70 deletions

diff --git a/bucket-lines-prob/bucket-lines-prob.go b/bucket-lines-prob/bucket-lines-prob.go
deleted file mode 100644
index 728268d..0000000
--- a/bucket-lines-prob/bucket-lines-prob.go
+++ /dev/null
@@ -1,56 +0,0 @@
-package main
-
-import (
-	"flag"
-	"fmt"
-	"log"
-	"os"
-
-	"git.rescribe.xyz/testingtools/parse"
-	"git.rescribe.xyz/testingtools/parse/prob"
-)
-
-func main() {
-	b := parse.BucketSpecs{
-		// minimum confidence, name
-		{ 0, "bad" },
-		{ 0.95, "95to98" },
-		{ 0.98, "98plus" },
-	}
-
-	flag.Usage = func() {
-		fmt.Fprintf(os.Stderr, "Usage: bucket-lines-prob [-d dir] prob1 [prob2] [...]\n")
-		fmt.Fprintf(os.Stderr, "Copies image-text line pairs into different directories according\n")
-		fmt.Fprintf(os.Stderr, "to the average character probability for the line.\n")
-		fmt.Fprintf(os.Stderr, "This uses the .prob files generated by ocropy-rpred's --probabilities\n")
-		fmt.Fprintf(os.Stderr, "option, which it assumes will be in the same directory as the line's\n")
-		fmt.Fprintf(os.Stderr, "image and text files.\n")
-		flag.PrintDefaults()
-	}
-	dir := flag.String("d", "buckets", "Directory to store the buckets")
-	flag.Parse()
-	if flag.NArg() < 1 {
-		flag.Usage()
-		os.Exit(1)
-	}
-
-	lines := make(parse.LineDetails, 0)
-
-	for _, f := range flag.Args() {
-		newlines, err := prob.GetLineDetails(f)
-		if err != nil {
-			log.Fatal(err)
-		}
-
-                for _, l := range newlines {
-                        lines = append(lines, l)
-                }
-	}
-
-	stats, err := parse.BucketUp(lines, b, *dir)
-	if err != nil {
-		log.Fatal(err)
-	}
-
-	parse.PrintBucketStats(os.Stdout, stats)
-}
diff --git a/bucket-lines-hocr/bucket-lines-hocr.go b/bucket-lines/bucket-lines.go
index b35c824..ad73fcd 100644
--- a/bucket-lines-hocr/bucket-lines-hocr.go
+++ b/bucket-lines/bucket-lines.go
@@ -1,7 +1,5 @@
 package main
 
-// TODO: merge with -prob, using filename extension to determine what to do for each file
-
 import (
 	"flag"
 	"fmt"
@@ -14,9 +12,11 @@ import (
 
 	"git.rescribe.xyz/testingtools/parse"
 	"git.rescribe.xyz/testingtools/parse/hocr"
+	"git.rescribe.xyz/testingtools/parse/prob"
 )
 
-func detailsFromFile(f string) (parse.LineDetails, error) {
+// TODO: maybe move this into hocr.go
+func detailsFromHocr(f string) (parse.LineDetails, error) {
 	var newlines parse.LineDetails
 
 	file, err := ioutil.ReadFile(f)
@@ -45,6 +45,9 @@ func detailsFromFile(f string) (parse.LineDetails, error) {
 }
 
 func main() {
+	// TODO: Allow different specs to be used for .prob vs .hocr. Do this
+	//       by adding a field to LineDetails that is linked to a named
+	//       BucketSpecs.
 	b := parse.BucketSpecs{
 		// minimum confidence, name
 		{ 0, "bad" },
@@ -53,13 +56,15 @@ func main() {
 	}
 
 	flag.Usage = func() {
-		fmt.Fprintf(os.Stderr, "Usage: bucket-lines-hocr [-d dir] hocr1 [hocr2] [...]\n")
+		fmt.Fprintf(os.Stderr, "Usage: bucket-lines [-d dir] [hocr1] [prob1] [hocr2] [...]\n")
 		fmt.Fprintf(os.Stderr, "Copies image-text line pairs into different directories according\n")
-		fmt.Fprintf(os.Stderr, "to the average character probability for the line.\n")
-		fmt.Fprintf(os.Stderr, "This uses the x_wconf data in .hocr files, which it assumes will be.\n")
-		fmt.Fprintf(os.Stderr, "in the same directory as the line's image and text files. It can\n")
-		fmt.Fprintf(os.Stderr, "handle hocr where each character is tagged separately and hocr where\n")
-		fmt.Fprintf(os.Stderr, "only whole words are tagged.\n")
+		fmt.Fprintf(os.Stderr, "to the average character probability for the line.\n\n")
+		fmt.Fprintf(os.Stderr, "Both .hocr and .prob files can be processed.\n\n")
+		fmt.Fprintf(os.Stderr, "For .hocr files, the x_wconf data is used to calculate confidence.\n\n")
+		fmt.Fprintf(os.Stderr, "The .prob files are generated using ocropy-rpred's --probabilities\n")
+		fmt.Fprintf(os.Stderr, "option.\n\n")
+		fmt.Fprintf(os.Stderr, "The .prob and .hocr files are assumed to be in the same directory\n")
+		fmt.Fprintf(os.Stderr, "as the line's image and text files.\n\n")
 		flag.PrintDefaults()
 	}
 	dir := flag.String("d", "buckets", "Directory to store the buckets")
@@ -69,10 +74,19 @@ func main() {
 		os.Exit(1)
 	}
 
+	var err error
 	lines := make(parse.LineDetails, 0)
 
 	for _, f := range flag.Args() {
-		newlines, err := detailsFromFile(f)
+		var newlines parse.LineDetails
+		switch ext := filepath.Ext(f); ext {
+			case ".prob":
+				newlines, err = prob.GetLineDetails(f)
+			case ".hocr":
+				newlines, err = detailsFromHocr(f)
+			default:
+				log.Printf("Skipping file '%s' as it isn't a .prob or .hocr\n", f)
+		}
 		if err != nil {
 			log.Fatal(err)
 		}
diff --git a/parse/hocr/hocr.go b/parse/hocr/hocr.go
index f7cac05..c03b73a 100644
--- a/parse/hocr/hocr.go
+++ b/parse/hocr/hocr.go
@@ -1,9 +1,10 @@
 package hocr
 
-// TODO: consider making GetLineDetails() a function of Hocr, so could do a
-//       similar thing with prob format files too, and then fire them both
-//       off a generic interface, potentially.
-// TODO: Parse line name to zero pad line numbers, so they come out in the correct order
+// TODO: Parse line name to zero pad line numbers, so they can
+//       be sorted easily
+// TODO: have same filename format as .prob uses, so include base
+//       dirname, and don't include line numbers if there's only
+//       one line in the hocr
 
 import (
 	"encoding/xml"