Generate preview image for PDF

6 files changed, 77 insertions, 4 deletions

diff --git a/CHANGES-DEV.md b/CHANGES-DEV.md
index b5807eeb4..0d21815f1 100644
--- a/CHANGES-DEV.md
+++ b/CHANGES-DEV.md
@@ -25,12 +25,13 @@ Ishtar develop changelog
 - Search:
   - document: operation year, operation type
   - context record - by operation name
-- commands:
+- Commands:
   - ishtar_maintenance_task: fix operation missing parcels
 - Tables:
   - operation - scientific column
-Sheet:
+- Sheet:
   - organization: sort member by alphabetic
+- Generate preview image for PDF
 
 
 ### Bugs ###
diff --git a/example_project/settings.py b/example_project/settings.py
index acf127121..53f15cc38 100644
--- a/example_project/settings.py
+++ b/example_project/settings.py
@@ -294,6 +294,7 @@ MAX_UPLOAD_SIZE = 100  # in Mo
 
 # path to the "dot" program to generate graph
 DOT_BINARY = "/usr/bin/dot"
+PDFTOPPM_BINARY = ""
 
 TEST_RUNNER = "ishtar_common.tests.ManagedModelTestRunner"
 SELENIUM_TEST = False
diff --git a/ishtar_common/forms_common.py b/ishtar_common/forms_common.py
index 572c1b70a..57efa72cd 100644
--- a/ishtar_common/forms_common.py
+++ b/ishtar_common/forms_common.py
@@ -1716,7 +1716,7 @@ class DocumentForm(forms.ModelForm, CustomForm, ManageOldType):
 
     def clean_source_page_range(self):
         value = self.cleaned_data.get("source_page_range", None).replace(" ", "")
-        if value and not re.match(r"^(\d+[-;]*\d)+$", value):
+        if value and not re.match(r"^(\d+[-;]*\d*)+$", value):
             raise forms.ValidationError(_("Incorrect page range."))
         return value
 
diff --git a/ishtar_common/models.py b/ishtar_common/models.py
index c46aad6c1..a133afead 100644
--- a/ishtar_common/models.py
+++ b/ishtar_common/models.py
@@ -81,6 +81,7 @@ from ishtar_common.utils import (
     get_image_path,
     serialize_args_for_tasks,
     task,
+    generate_pdf_preview,
 )
 from ishtar_common.utils_secretary import IshtarSecretaryRenderer
 
@@ -4615,7 +4616,14 @@ class Document(
                 self.save(no_path_change=True)
 
 
-post_save.connect(cached_label_changed, sender=Document)
+def document_changed(sender, **kwargs):
+    cached_label_changed(sender, **kwargs)
+    if not settings.PDFTOPPM_BINARY or not kwargs.get("instance"):
+        return
+    generate_pdf_preview(kwargs.get("instance"))
+
+
+post_save.connect(document_changed, sender=Document)
 
 
 class OperationType(GeneralType):
diff --git a/ishtar_common/tests.py b/ishtar_common/tests.py
index 6b200ef39..0c790bdd5 100644
--- a/ishtar_common/tests.py
+++ b/ishtar_common/tests.py
@@ -3656,6 +3656,24 @@ class DocumentTest(TestCase):
         doc.operations.add(self.ope1)
         self.assertEqual(doc.operations.count(), 0)
 
+    def test_generate_pdftoppm(self):
+        if not settings.PDFTOPPM_BINARY:
+            return
+        pdf_path = os.path.join(
+            settings.ROOT_PATH, "..", "ishtar_common", "tests", "simple.pdf"
+        )
+        doc = models.Document.objects.create(
+            title="Document",
+            associated_file=SimpleUploadedFile(
+                name="simple.pdf",
+                content=open(pdf_path, "rb").read(),
+                content_type="application/pdf",
+            )
+        )
+        doc.operations.add(self.ope1)
+        doc = models.Document.objects.get(id=doc.pk)
+        self.assertTrue(doc.image.path)
+
     def test_create_with_parent(self):
         doc = models.Document.objects.create(title="Parent document")
         doc.operations.add(self.ope1)
diff --git a/ishtar_common/utils.py b/ishtar_common/utils.py
index 173338b4f..bfb5d0e32 100644
--- a/ishtar_common/utils.py
+++ b/ishtar_common/utils.py
@@ -2117,3 +2117,48 @@ class IshtarFileSystemStorage(FileSystemStorage):
             if not os.path.exists(os.readlink(path_name)):
                 os.remove(path_name)
         return os.path.exists(path_name)
+
+
+def generate_pdf_preview(item, save=True, tempdir=None, page_number=None):
+    if not settings.PDFTOPPM_BINARY:
+        return
+    preview_tmp_name = None
+    if not item.image and item.source and item.source_page_range:
+        tempdir = tempfile.mkdtemp("-ishtarppmtopdf")
+        try:
+            page_number = int(
+                item.source_page_range.split("-")[0].split(";")[0].split(",")[0])
+        except ValueError:
+            page_number = 1
+        returned = generate_pdf_preview(item.source, save=False, tempdir=tempdir,
+                                        page_number=page_number)
+        if not returned:
+            return
+        tempdir, preview_tmp_name = returned
+    elif (not page_number and item.image) \
+            or not item.associated_file \
+            or not item.associated_file.path \
+            or not item.associated_file.path.lower().endswith(".pdf"):
+        return
+    if not page_number:
+        page_number = 1
+
+    if not tempdir:
+        tempdir = tempfile.mkdtemp("-ishtarppmtopdf")
+    if not preview_tmp_name:
+        preview_tmp_name = tempdir + os.path.sep + "preview"
+        args = (settings.PDFTOPPM_BINARY, "-singlefile", "-jpeg", "-f", str(page_number),
+                item.associated_file.path, preview_tmp_name)
+
+        try:
+            popen = subprocess.Popen(args)
+            popen.wait(timeout=5)
+        except subprocess.SubprocessError:
+            return
+    if not save:
+        return tempdir, preview_tmp_name
+
+    with open(preview_tmp_name + ".jpg", "rb") as preview:
+        django_file = File(preview)
+        getattr(item, "image").save("page.jpg", django_file, save=True)
+    shutil.rmtree(tempdir)