Visual FoxPro Requires Chilkat v11.7.0+

Assess PDF Text Quality Before Extracting

See more PDF Text Extraction Examples

Demonstrates the "assess" operation, which returns only the quality report: whether the text can be trusted, which pages are scanned images that need OCR, and how each page's text was found. Use it to triage documents before extracting. (Chilkat does not perform OCR.) Visual FoxPro

LOCAL lnSuccess
LOCAL loPdf
LOCAL loJson
LOCAL loJsonOut
LOCAL j
LOCAL lnCount_j
LOCAL i
LOCAL lnCount_i

lnSuccess = 0

*  This example requires the Chilkat API to have been previously unlocked.
*  See Global Unlock Sample for sample code.

*  The PDF used in this example (page 2 is a scanned image with no text) is available at mixed_text_image.pdf

loPdf = CreateObject('Chilkat.Pdf')
lnSuccess = loPdf.LoadFile("qa_data/pdf/mixed_text_image.pdf")
IF (lnSuccess = 0) THEN
    ? loPdf.LastErrorText
    RELEASE loPdf
    CANCEL
ENDIF

loJson = CreateObject('Chilkat.JsonObject')
loJsonOut = CreateObject('Chilkat.JsonObject')

*  "assess" computes the quality report without producing any text.
*  It is fast, and it is always allowed, even for an encrypted PDF that forbids copying.
lnSuccess = loPdf.ExtractText("assess",loJson,loJsonOut)
IF (lnSuccess = 0) THEN
    ? loPdf.LastErrorText
    RELEASE loPdf
    RELEASE loJson
    RELEASE loJsonOut
    CANCEL
ENDIF

*  The overall rating is the worst rating among the pages that have text:
*  "good", "fair", "poor", or "none" when there is no text at all.
? "Rating: " + loJsonOut.StringOf("quality.rating")

*  Is the PDF tagged (does it have a structure tree)?
? "Tagged: " + STR(loJsonOut.BoolOf("quality.tagged"))

*  The pages that are images only and would need OCR, such as "2" or "4,9".
? "Pages needing OCR: " + loJsonOut.StringOf("quality.pagesNeedingOcr")

*  The per-page report.

i = 0
lnCount_i = loJsonOut.SizeOfArray("quality.pages")
DO WHILE i < lnCount_i
    loJsonOut.I = i
    *  textSource is "text" (real text), "ocrLayer" (invisible text over an image),
    *  "imageOnly" (needs OCR), "mixed", or "empty".
    ? "Page " + STR(loJsonOut.IntOf("quality.pages[i].page")) + ": rating " + loJsonOut.StringOf("quality.pages[i].rating") + ", textSource " + loJsonOut.StringOf("quality.pages[i].textSource")
    *  Each issue has a machine-readable code and a sentence.
    j = 0
    lnCount_j = loJsonOut.SizeOfArray("quality.pages[i].issues")
    DO WHILE j < lnCount_j
        loJsonOut.J = j
        ? "    " + loJsonOut.StringOf("quality.pages[i].issues[j].code") + ": " + loJsonOut.StringOf("quality.pages[i].issues[j].message")
        j = j + 1
    ENDDO
    i = i + 1
ENDDO

*  Decide what to do with the document:
IF (loJsonOut.StringOf("quality.rating") = "none") THEN
    ? "No extractable text; the whole document needs OCR."
ELSE
    IF (loJsonOut.StringOf("quality.pagesNeedingOcr") <> "") THEN
        ? "Some pages need OCR; the remaining pages can be extracted with the text or markdown operation."
    ELSE
        ? "All pages can be extracted."
    ENDIF

ENDIF

*  Sample output:

*  Rating: good
*  Tagged: false
*  Pages needing OCR: 2
*  Page 1: rating good, textSource text
*  Page 2: rating none, textSource imageOnly
*      imageOnly: No text; page is a scanned image
*  Page 3: rating good, textSource text
*  Some pages need OCR; the remaining pages can be extracted with the text or markdown operation.

RELEASE loPdf
RELEASE loJson
RELEASE loJsonOut