contributes towards sajari#18

pdf.go

@@ -4,7 +4,10 @@ import (
 	"fmt"
 	"io"
 	"log"
+	"os"
 	"os/exec"
+	"path/filepath"
+	"strconv"
 	"strings"
 	"time"
 )
@@ -18,7 +21,7 @@ func ConvertPDF(r io.Reader) (string, map[string]string, error) {
 	defer f.Done()
 
 	// Meta data
-	mc := make(chan map[string]string, 1)
+	mc := make(chan map[string]string)
 	go func() {
 		meta := make(map[string]string)
 		metaStr, err := exec.Command("pdfinfo", f.Name()).Output()
@@ -57,6 +60,77 @@ func ConvertPDF(r io.Reader) (string, map[string]string, error) {
 			// TODO: Remove this.
 			log.Println("pdftotext:", err)
 		}
+		if len(body) == 0 {
+
+			// grab the metadata, mitm-style
+			meta := <-mc
+			go func() { mc <- meta }()
+
+			pages, err := strconv.Atoi(meta["Pages"])
+			if err != nil {
+				log.Printf("failed to get number of pages from '%s': %v", meta["pages"], err)
+				return
+			}
+
+			dir, err := filepath.Abs(filepath.Dir(os.Args[0]))
+			if err != nil {
+				log.Fatal(err)
+			}
+
+			for page := 0; page < pages; page++ {
+				tmpImageFileName := fmt.Sprintf("%s-%d.jpg", f.Name(), page)
+				args := []string{
+					fmt.Sprintf("-dFirstPage=%d", page+1),
+					fmt.Sprintf("-dLastPage=%d", page+1),
+					"-dNOGC",
+					// "-dQUIET",
+					// "-dBandHeight=100",
+					// "-dBandBufferSpace=500000000",
+					// "-dBufferSpace=1000000000",
+					// "-sBandListStorage=memory",
+					// "-dNumRenderingThreads=2",
+					"-sDEVICE=jpeg",
+					"-r300",
+					"-o", tmpImageFileName,
+					"-f", filepath.Join(dir, f.Name()),
+				}
+
+				_, err := exec.Command("gs", args...).Output()
+				if err != nil {
+					log.Printf("failed to execute gs: %v", err)
+					return
+				}
+
+				// stdout, err := cmd.StdoutPipe()
+				// if err != nil {
+				// 	log.Printf("failed to get stdout from gs call: %v", err)
+				// 	return
+				// }
+
+				tmpImageFile, err := os.Open(tmpImageFileName)
+				if err != nil {
+					log.Printf("failed to open temp image file: %v", err)
+					return
+				}
+				defer func() {
+					err := tmpImageFile.Close()
+					if err != nil {
+						log.Printf("failed to close tmpImageFile: %v", err)
+					}
+					err = os.Remove(tmpImageFileName)
+					if err != nil {
+						log.Printf("failed to remove tmpImageFileName: %v", err)
+					}
+				}()
+
+				result, _, err := ConvertImage(tmpImageFile)
+				if err != nil {
+					log.Printf("failed to run ConvertImage on gs output: %v", err)
+					return
+				}
+				bc <- result
+			}
+		}
 		bc <- string(body)
 	}()