MLCP导入时调用Transform模块遇前缀无命名空间绑定错误

报错信息 #

23/01/13 15:10:37 INFO contentpump.FileAndDirectoryInputFormat: Total input paths to process : 1
23/01/13 15:10:38 ERROR contentpump.TransformWriter: Batch 361007023.0: Document failed permanently: /opt/mlcp-10.0.9-bin/mlcp-10.0.9/bin/data/EDI_22000041.xml
23/01/13 15:10:38 WARN contentpump.TransformWriter: <error:format-string xmlns:error="http://marklogic.com/xdmp/error" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">XDMP-UNBPRFX: (err:XPST0081) Prefix doc has no namespace binding</error:format-string>
23/01/13 15:10:38 INFO contentpump.LocalJobRunner:  completed 100%
23/01/13 15:10:38 INFO contentpump.LocalJobRunner: com.marklogic.mapreduce.MarkLogicCounter:
23/01/13 15:10:38 INFO contentpump.LocalJobRunner: INPUT_RECORDS: 1
23/01/13 15:10:38 INFO contentpump.LocalJobRunner: OUTPUT_RECORDS: 1
23/01/13 15:10:38 INFO contentpump.LocalJobRunner: OUTPUT_RECORDS_COMMITTED: 0
23/01/13 15:10:38 INFO contentpump.LocalJobRunner: OUTPUT_RECORDS_FAILED: 1

Transform模块代码（原错误版本） #

xquery version "1.0-ml";
module namespace doc = "http://marklogic.com/example";

declare function doc:uri(
  $content as map:map,
  $context as map:map
) as map:map*
{
  let $changeduri := map:get($content, "value")/EDILogFile/DateCreated
  return
    (
      map:put(
          $content, "/EDI"||"_"||$changeduri||".xml")
    )
};

加载Transform模块的XQuery代码（原版本） #

xquery version "1.0-ml";
xdmp:document-load("/opt/urichange.xqy",
    <options xmlns="xdmp:document-load">
      <uri>/opt/urichange.xqy</uri>
      <repair>none</repair>
      <permissions>{xdmp:default-permissions()}</permissions>
    </options>)

待导入文件内容 #

<?xml version="1.0" encoding="UTF-8"?>
    <EDILogFile xsi:noNamespaceSchemaLocation="file:///Projects/test/schema/LogFile.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">
        <ID>****</ID>
        <SystemRef>22000041</SystemRef>
        <DateCreated>2022-09-06T19:07:46.3492849+01:00</DateCreated>
        <TimeSaved>240</TimeSaved>
        <Production>true</Production>
        <Partner>test</Partner>
        <MessageType>test</MessageType>
        <Fail>false</Fail>
        <ManyReasons/>
        <SubmissionUser>test</SubmissionUser>
        <InternalBusinessUnit>test</InternalBusinessUnit>
        <Direction>Inbound</Direction>
    </EDILogFile>

问题排查与解决 #

1. 命名空间绑定缺失 #

报错核心是前缀doc无命名空间绑定，原因是MLCP调用Transform模块时，未指定模块对应的命名空间。Transform模块定义了module namespace doc = "http://marklogic.com/example";，因此MLCP命令必须添加-transform_namespace参数指定该命名空间，否则无法识别doc:uri函数。

2. Transform函数参数错误 #

原代码中map:put用法错误：map:put需要三个参数（map对象、键名、值），修改文档URI应更新$content中的uri键，而非直接传入URI字符串作为第二个参数。

3. 节点值处理不当 #

$changeduri是XML节点，直接拼接会包含节点标签结构，需提取其字符串值用于URI拼接。

4. 模块加载路径错误 #

加载Transform模块时，<uri>指定的是本地路径/opt/urichange.xqy，但MLCP调用的是MarkLogic数据库内的模块，因此需将模块加载到数据库内的合理路径（如/urichange.xqy）。

修正后的代码与命令 #

修正后的Transform模块 #

xquery version "1.0-ml";
module namespace doc = "http://marklogic.com/example";

declare function doc:uri(
  $content as map:map,
  $context as map:map
) as map:map*
{
  let $date-created := string(map:get($content, "value")/EDILogFile/DateCreated)
  let $new-uri := "/EDI_" || $date-created || ".xml"
  return map:put($content, "uri", $new-uri)
};

修正后的模块加载XQuery #

xquery version "1.0-ml";
xdmp:document-load("/opt/urichange.xqy",
    <options xmlns="xdmp:document-load">
      <uri>/urichange.xqy</uri> <!-- 数据库内的模块URI -->
      <repair>none</repair>
      <permissions>{xdmp:default-permissions()}</permissions>
    </options>)

正确的MLCP导入命令示例 #

mlcp import -host localhost -port 8000 -username admin -password admin \
-input_file_path /opt/mlcp-10.0.9-bin/mlcp-10.0.9/bin/data/EDI_22000041.xml \
-transform_module /urichange.xqy \
-transform_namespace http://marklogic.com/example

内容的提问来源于stack exchange，提问作者Rashmita Purkayastha