From 534c7b4bf7052c103d94e5761631979e23411e39 Mon Sep 17 00:00:00 2001
From: pubpub-zz <4083478+pubpub-zz@users.noreply.github.com>
Date: Sun, 30 Jul 2023 17:28:55 +0200
Subject: [PATCH] ENH: Process /uniHHHH for text_extract (#2043)

`/uniHHHH` (H is a hexadecimal) glyphs seems to be generated in LaTeX but is ok for other characters

This was mentioned in #2016 / #2038
---
 pypdf/_cmap.py           | 15 ++++++++++++---
 tests/test_cmap.py       | 10 ++++++++++
 tests/test_encryption.py | 22 +++++++++++++++++-----
 3 files changed, 39 insertions(+), 8 deletions(-)

diff --git a/pypdf/_cmap.py b/pypdf/_cmap.py
index 14c1e229c..6400f89b1 100644
--- a/pypdf/_cmap.py
+++ b/pypdf/_cmap.py
@@ -461,10 +461,19 @@ def type1_alternative(
                 continue
             try:
                 i = int(words[1])
-                v = adobe_glyphs[words[2].decode()]
-            except (ValueError, KeyError):
+            except ValueError:  # pragma: no cover
                 continue
-            if v == " ":
+            try:
+                v = adobe_glyphs[words[2].decode()]
+            except KeyError:
+                if words[2].startswith(b"/uni"):
+                    try:
+                        v = chr(int(words[2][4:], 16))
+                    except ValueError:  # pragma: no cover
+                        continue
+                else:
+                    continue
+            if words[2].decode() == b" ":
                 space_code = i
             map_dict[chr(i)] = v
             int_entry.append(i)
diff --git a/tests/test_cmap.py b/tests/test_cmap.py
index f74da326d..6e7448651 100644
--- a/tests/test_cmap.py
+++ b/tests/test_cmap.py
@@ -179,3 +179,13 @@ def test_latex():
     for pat in ("α", "β", "γ", "ϕ", "φ", "ℏ", "∫", "∂", "·", "×"):
         assert pat in txt
     # actually the ϕ and φ seems to be crossed in latex
+
+
+@pytest.mark.enable_socket()
+def test_unixxx_glyphs():
+    url = "https://arxiv.org/pdf/2201.00021.pdf"
+    name = "unixxx_glyphs.pdf"
+    reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
+    txt = reader.pages[0].extract_text()  # no error
+    for pat in ("闫耀庭", "龚龑", "张江水", "1′′.2"):
+        assert pat in txt
diff --git a/tests/test_encryption.py b/tests/test_encryption.py
index ff33d2121..6641977a4 100644
--- a/tests/test_encryption.py
+++ b/tests/test_encryption.py
@@ -125,7 +125,10 @@ def test_encryption(name, requires_pycryptodome):
         ("r6-both-passwords.pdf", "foo", "bar"),
     ],
 )
-@pytest.mark.skipif(not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY, reason="No pycryptodome / cryptography")
+@pytest.mark.skipif(
+    not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY,
+    reason="No pycryptodome / cryptography",
+)
 def test_pdf_with_both_passwords(name, user_passwd, owner_passwd):
     """
     PDFs with both user and owner passwords are handled correctly.
@@ -151,7 +154,10 @@ def test_pdf_with_both_passwords(name, user_passwd, owner_passwd):
         ("crazyones-encrypted-256.pdf", b"password"),
     ],
 )
-@pytest.mark.skipif(not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY, reason="No pycryptodome / cryptography")
+@pytest.mark.skipif(
+    not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY,
+    reason="No pycryptodome / cryptography",
+)
 def test_read_page_from_encrypted_file_aes_256(pdffile, password):
     """
     A page can be read from an encrypted.
@@ -176,7 +182,10 @@ def test_read_page_from_encrypted_file_aes_256(pdffile, password):
         ),
     ],
 )
-@pytest.mark.skipif(not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY, reason="No pycryptodome / cryptography")
+@pytest.mark.skipif(
+    not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY,
+    reason="No pycryptodome / cryptography",
+)
 @pytest.mark.filterwarnings("ignore::DeprecationWarning")
 def test_merge_encrypted_pdfs(names):
     """Encrypted PDFs can be merged after decryption."""
@@ -193,7 +202,7 @@ def test_merge_encrypted_pdfs(names):
 
 @pytest.mark.skipif(
     HAS_CRYPTOGRAPHY,
-    reason="Limitations of cryptography. see https://github.com/pyca/cryptography/issues/2494"
+    reason="Limitations of cryptography. see https://github.com/pyca/cryptography/issues/2494",
 )
 @pytest.mark.parametrize(
     "cryptcls",
@@ -346,7 +355,10 @@ def test_pdf_encrypt_multiple(pdf_file_path, count):
     assert text0 == text1
 
 
-@pytest.mark.skipif(not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY, reason="No pycryptodome / cryptography")
+@pytest.mark.skipif(
+    not HAS_PYCRYPTODOME and not HAS_CRYPTOGRAPHY,
+    reason="No pycryptodome / cryptography",
+)
 def test_aes_decrypt_corrupted_data():
     """Just for robustness"""
     aes = CryptAES(secrets.token_bytes(16))