Merge pull request #54097 from bharatnc/ncb/decode-html-component

add function decodeHTMLComponent
ClickHouse · Sep 4, 2023 · ec628ee · ec628ee
2 parents 476e15c + 8ecbdd2
commit ec628ee
Show file tree

Hide file tree

Showing 11 changed files with 20,511 additions and 0 deletions.
diff --git a/docs/en/sql-reference/functions/string-functions.md b/docs/en/sql-reference/functions/string-functions.md
@@ -1230,6 +1230,42 @@ Result:
 < Σ >
 ```
 
+## decodeHTMLComponent
+
+Un-escapes substrings with special meaning in HTML. For example: `&hbar;` `&gt;` `&diamondsuit;` `&heartsuit;` `&lt;` etc.
+
+This function also replaces numeric character references with Unicode characters. Both decimal (like `&#10003;`) and hexadecimal (`&#x2713;`) forms are supported.
+
+**Syntax**
+
+``` sql
+decodeHTMComponent(x)
+```
+
+**Arguments**
+
+- `x` — An input string. [String](../../sql-reference/data-types/string.md).
+
+**Returned value**
+
+- The un-escaped string.
+
+Type: [String](../../sql-reference/data-types/string.md).
+
+**Example**
+
+``` sql
+SELECT decodeHTMLComponent(''CH');
+SELECT decodeHMLComponent('I&heartsuit;ClickHouse');
+```
+
+Result:
+
+```result
+'CH'
+I♥ClickHouse'
+```
+
 ## extractTextFromHTML
 
 This function extracts plain text from HTML or XHTML.

diff --git a/src/Functions/CMakeLists.txt b/src/Functions/CMakeLists.txt
@@ -124,6 +124,27 @@ if (ENABLE_FUZZING)
     add_compile_definitions(FUZZING_MODE=1)
 endif ()
 
+if (USE_GPERF)
+    # Only for regenerating
+    add_custom_target(generate-html-char-ref-gperf ./HTMLCharacterReference.sh
+            SOURCES ./HTMLCharacterReference.sh
+            WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+            BYPRODUCTS "${CMAKE_CURRENT_SOURCE_DIR}/HTMLCharacterReference.gperf"
+            )
+    add_custom_target(generate-html-char-ref ${GPERF} -t HTMLCharacterReference.gperf --output-file=HTMLCharacterReference.generated.cpp
+            && clang-format -i HTMLCharacterReference.generated.cpp
+            # for clang-tidy, since string.h is deprecated
+            && sed -i 's/\#include <string.h>/\#include <cstring>/g' HTMLCharacterReference.generated.cpp
+            SOURCES HTMLCharacterReference.gperf
+            WORKING_DIRECTORY ${CMAKE_CURRENT_SOURCE_DIR}
+            )
+    add_dependencies(generate-html-char-ref generate-html-char-ref-gperf)
+    if (NOT TARGET generate-source)
+        add_custom_target(generate-source)
+    endif ()
+    add_dependencies(generate-source generate-html-char-ref)
+endif ()
+
 target_link_libraries(clickhouse_functions_obj PUBLIC ${PUBLIC_LIBS} PRIVATE ${PRIVATE_LIBS})
 
 # Used to forward the linking information to the final binaries such as clickhouse / unit_tests_dbms,