feat(table): build a catalog-free Table from a portable TableDescriptor

.licenserc.yaml

@@ -27,6 +27,7 @@ header:
     - ".gitattributes"
     - ".github/PULL_REQUEST_TEMPLATE.md"
     - "crates/paimon/tests/**/*.json"
+    - "crates/paimon/src/table/goldens/*.json"
     - "crates/paimon/testdata/**"
     - "third-party-licenses/openssl-1.1.1.LICENSE"
     - "**/go.sum"

bindings/c/src/table.rs

@@ -17,6 +17,7 @@
 
 use std::collections::HashMap;
 use std::ffi::c_void;
+use std::ptr;
 
 use arrow_array::ffi::{FFI_ArrowArray, FFI_ArrowSchema};
 use arrow_array::{Array, StructArray};
@@ -27,8 +28,9 @@ use paimon::Plan;
 
 use crate::error::{check_non_null, paimon_error, validate_cstr, PaimonErrorCode};
 use crate::result::{
-    paimon_result_new_read, paimon_result_next_batch, paimon_result_plan, paimon_result_predicate,
-    paimon_result_read_builder, paimon_result_record_batch_reader, paimon_result_table_scan,
+    paimon_result_get_table, paimon_result_new_read, paimon_result_next_batch, paimon_result_plan,
+    paimon_result_predicate, paimon_result_read_builder, paimon_result_record_batch_reader,
+    paimon_result_table_scan,
 };
 use crate::runtime;
 use crate::types::*;
@@ -61,12 +63,96 @@ unsafe fn box_table_read_state(state: TableReadState) -> *mut paimon_table_read
 /// Free a paimon_table.
 ///
 /// # Safety
-/// Only call with a table returned from `paimon_catalog_get_table`.
+/// Only call with a table returned from `paimon_catalog_get_table` or
+/// `paimon_table_from_descriptor`.
 #[no_mangle]
 pub unsafe extern "C" fn paimon_table_free(table: *mut paimon_table) {
     free_table_wrapper(table, |t| t.inner);
 }
 
+/// Build a catalog-free `paimon_table` (intended for scan/read) from an
+/// FE-provided `TableDescriptor` JSON, without any catalog lookup.
+///
+/// `descriptor_json` is a NUL-terminated UTF-8 JSON string (see the Rust
+/// `paimon::table::TableDescriptor`). `options` is an array of `options_len`
+/// storage/runtime `paimon_option` values (credentials, endpoints), or null
+/// when `options_len` is 0; they configure `FileIO` only and never change table
+/// semantics. The returned table is intended for scan/read.
+///
+/// On success `table` is non-null and `error` is null; free it with
+/// `paimon_table_free`. On failure `table` is null and `error` is non-null.
+///
+/// # Safety
+/// `descriptor_json` must be a valid C string. `options` must point to
+/// `options_len` valid `paimon_option` values, or be null when `options_len` is 0.
+#[no_mangle]
+pub unsafe extern "C" fn paimon_table_from_descriptor(
+    descriptor_json: *const std::ffi::c_char,
+    options: *const paimon_option,
+    options_len: usize,
+) -> paimon_result_get_table {
+    let json = match validate_cstr(descriptor_json, "descriptor_json") {
+        Ok(s) => s,
+        Err(e) => {
+            return paimon_result_get_table {
+                table: ptr::null_mut(),
+                error: e,
+            }
+        }
+    };
+
+    if options.is_null() && options_len > 0 {
+        return paimon_result_get_table {
+            table: ptr::null_mut(),
+            error: paimon_error::new(
+                PaimonErrorCode::InvalidInput,
+                "null options pointer with non-zero length".to_string(),
+            ),
+        };
+    }
+    let mut opts: HashMap<String, String> = HashMap::new();
+    if options_len > 0 {
+        let options_slice = std::slice::from_raw_parts(options, options_len);
+        for opt in options_slice {
+            let key = match validate_cstr(opt.key, "option key") {
+                Ok(s) => s,
+                Err(e) => {
+                    return paimon_result_get_table {
+                        table: ptr::null_mut(),
+                        error: e,
+                    }
+                }
+            };
+            let value = match validate_cstr(opt.value, "option value") {
+                Ok(s) => s,
+                Err(e) => {
+                    return paimon_result_get_table {
+                        table: ptr::null_mut(),
+                        error: e,
+                    }
+                }
+            };
+            opts.insert(key, value);
+        }
+    }
+
+    match Table::from_descriptor_json(&json, opts) {
+        Ok(table) => {
+            let wrapper = Box::new(paimon_table {
+                inner: Box::into_raw(Box::new(table)) as *mut c_void,
+            });
+            paimon_result_get_table {
+                table: Box::into_raw(wrapper),
+                error: ptr::null_mut(),
+            }
+        }
+        Err(e) => paimon_result_get_table {
+            table: ptr::null_mut(),
+            error: paimon_error::from_paimon(e),
+        },
+    }
+}
+
 /// Time-travel selector option names, in the core's resolution priority order.
 const TIME_TRAVEL_SELECTORS: [&str; 4] = [
     "scan.timestamp-millis",
@@ -1787,6 +1873,16 @@ const _: unsafe extern "C" fn(
 const _: unsafe extern "C" fn(*const u8, usize) -> paimon_result_plan =
     paimon_plan_from_split_bytes;
 
+// Descriptor table constructor ABI signature guard. Pins the catalog-free table
+// constructor so an accidental signature change fails to compile rather than
+// silently breaking header consumers. To add behavior, introduce a new symbol
+// instead of changing this one.
+const _: unsafe extern "C" fn(
+    *const std::ffi::c_char,
+    *const paimon_option,
+    usize,
+) -> paimon_result_get_table = paimon_table_from_descriptor;
+
 #[cfg(test)]
 mod tests {
     use super::*;

bindings/c/src/tests.rs

@@ -2474,3 +2474,55 @@ fn plan_from_split_bytes_rejects_garbage() {
     assert!(!r.error.is_null());
     unsafe { paimon_error_free(r.error) };
 }
+
+// A minimal but valid v1 TableDescriptor JSON (scalar schema; the descriptor
+// parser's complex-type coverage lives in the paimon crate's unit tests).
+const DESCRIPTOR_V1_JSON: &str = r#"{
+  "version": 1,
+  "path": "file:///tmp/paimon-c-test/db.db/t",
+  "database": "db",
+  "name": "t",
+  "tableSchema": {
+    "version": 3,
+    "id": 0,
+    "fields": [ { "id": 0, "name": "id", "type": "INT NOT NULL" } ],
+    "highestFieldId": 0,
+    "partitionKeys": [],
+    "primaryKeys": ["id"],
+    "options": {},
+    "timeMillis": 1
+  }
+}"#;
+
+#[test]
+fn table_from_descriptor_builds_table() {
+    let json = CString::new(DESCRIPTOR_V1_JSON).unwrap();
+    unsafe {
+        let result = paimon_table_from_descriptor(json.as_ptr(), ptr::null(), 0);
+        assert!(result.error.is_null(), "from_descriptor should not error");
+        assert!(!result.table.is_null(), "table handle must be non-null");
+        paimon_table_free(result.table);
+    }
+}
+
+#[test]
+fn table_from_descriptor_rejects_unknown_version() {
+    let json =
+        CString::new(DESCRIPTOR_V1_JSON.replace("\"version\": 1", "\"version\": 2")).unwrap();
+    unsafe {
+        let result = paimon_table_from_descriptor(json.as_ptr(), ptr::null(), 0);
+        assert!(!result.error.is_null(), "unknown version must error");
+        assert!(result.table.is_null());
+        paimon_error_free(result.error);
+    }
+}
+
+#[test]
+fn table_from_descriptor_rejects_null_json() {
+    unsafe {
+        let result = paimon_table_from_descriptor(ptr::null(), ptr::null(), 0);
+        assert!(!result.error.is_null(), "null json must error");
+        assert!(result.table.is_null());
+        paimon_error_free(result.error);
+    }
+}