fix #1220: parse step url with base url

hrp/parser.go

@@ -4,6 +4,7 @@ import (
 	builtinJSON "encoding/json"
 	"fmt"
 	"net/url"
+	"path"
 	"reflect"
 	"regexp"
 	"strings"
@@ -26,18 +27,29 @@ type Parser struct {
 }
 
 func buildURL(baseURL, stepURL string) string {
-	uConfig, err := url.Parse(baseURL)
+	uStep, err := url.Parse(stepURL)
 	if err != nil {
-		log.Error().Str("baseURL", baseURL).Err(err).Msg("[buildURL] parse baseURL failed")
+		log.Error().Str("stepURL", stepURL).Err(err).Msg("[buildURL] parse url failed")
 		return ""
 	}
 
-	uStep, err := uConfig.Parse(stepURL)
+	// step url is absolute url
+	if uStep.Host != "" {
+		return stepURL
+	}
+
+	// step url is relative, based on base url
+	uConfig, err := url.Parse(baseURL)
 	if err != nil {
-		log.Error().Str("stepURL", stepURL).Err(err).Msg("[buildURL] parse stepURL failed")
+		log.Error().Str("baseURL", baseURL).Err(err).Msg("[buildURL] parse url failed")
 		return ""
 	}
 
+	// merge url
+	uStep.Scheme = uConfig.Scheme
+	uStep.Host = uConfig.Host
+	uStep.Path = path.Join(uConfig.Path, uStep.Path)
+
 	// base url missed
 	return uStep.String()
 }

hrp/parser_test.go

@@ -11,25 +11,44 @@ import (
 
 func TestBuildURL(t *testing.T) {
 	var url string
+
 	url = buildURL("https://postman-echo.com", "/get")
-	if url != "https://postman-echo.com/get" {
+	if !assert.Equal(t, url, "https://postman-echo.com/get") {
-		t.Fatalf("buildURL error, %s != 'https://postman-echo.com/get'", url)
+		t.Fail()
+	}
+	url = buildURL("https://postman-echo.com", "get")
+	if !assert.Equal(t, url, "https://postman-echo.com/get") {
+		t.Fail()
+	}
+	url = buildURL("https://postman-echo.com/", "/get")
+	if !assert.Equal(t, url, "https://postman-echo.com/get") {
+		t.Fail()
 	}
 
 	url = buildURL("https://postman-echo.com/abc/", "/get?a=1&b=2")
-	if url != "https://postman-echo.com/get?a=1&b=2" {
+	if !assert.Equal(t, url, "https://postman-echo.com/abc/get?a=1&b=2") {
-		t.Fatalf("buildURL error, %s != 'https://postman-echo.com/get'", url)
+		t.Fail()
+	}
+	url = buildURL("https://postman-echo.com/abc", "get?a=1&b=2")
+	if !assert.Equal(t, url, "https://postman-echo.com/abc/get?a=1&b=2") {
+		t.Fail()
+	}
+
+	// omit query string in base url
+	url = buildURL("https://postman-echo.com/abc?x=6&y=9", "/get?a=1&b=2")
+	if !assert.Equal(t, url, "https://postman-echo.com/abc/get?a=1&b=2") {
+		t.Fail()
 	}
 
 	url = buildURL("", "https://postman-echo.com/get")
-	if url != "https://postman-echo.com/get" {
+	if !assert.Equal(t, url, "https://postman-echo.com/get") {
-		t.Fatalf("buildURL error, %s != 'https://postman-echo.com/get'", url)
+		t.Fail()
 	}
 
 	// notice: step request url > config base url
 	url = buildURL("https://postman-echo.com", "https://httpbin.org/get")
-	if url != "https://httpbin.org/get" {
+	if !assert.Equal(t, url, "https://httpbin.org/get") {
-		t.Fatalf("buildURL error, %s != 'https://httpbin.org/get'", url)
+		t.Fail()
 	}
 }
 

httprunner/parser.py

@@ -3,6 +3,7 @@ import builtins
 import os
 import re
 from typing import Any, Callable, Dict, List, Set, Text
+from urllib.parse import urljoin, urlparse
 
 from loguru import logger
 from sentry_sdk import capture_exception
@@ -10,8 +11,6 @@ from sentry_sdk import capture_exception
 from httprunner import exceptions, loader, utils
 from httprunner.models import FunctionsMapping, VariablesMapping
 
-absolute_http_url_regexp = re.compile(r"^https?://", re.I)
-
 # use $$ to escape $ notation
 dolloar_regex_compile = re.compile(r"\$\$")
 # variable notation, e.g. ${var} or $var
@@ -37,15 +36,25 @@ def parse_string_value(str_value: Text) -> Any:
         return str_value
 
 
-def build_url(base_url, path):
+def build_url(base_url, step_url):
     """ prepend url with base_url unless it's already an absolute URL """
-    if absolute_http_url_regexp.match(path):
+    o_step_url = urlparse(step_url)
-        return path
+    if o_step_url.netloc != "":
-    elif base_url:
+        # step url is absolute url
-        return "{}/{}".format(base_url.rstrip("/"), path.lstrip("/"))
+        return step_url
-    else:
+
+    # step url is relative, based on base url
+    o_base_url = urlparse(base_url)
+    if o_base_url.netloc == "":
+        # missed base url
         raise exceptions.ParamsError("base url missed!")
 
+    path = o_base_url.path.rstrip("/") + "/" + o_step_url.path.lstrip("/")
+    o_step_url = o_step_url._replace(scheme=o_base_url.scheme) \
+        ._replace(netloc=o_base_url.netloc) \
+        ._replace(path=path)
+    return o_step_url.geturl()
+
 
 def regex_findall_variables(raw_string: Text) -> List[Text]:
     """ extract all variable names from content, which is in format $variable

httprunner/parser_test.py

@@ -3,11 +3,36 @@ import time
 import unittest
 
 from httprunner import parser
-from httprunner.exceptions import VariableNotFound, FunctionNotFound
+from httprunner.exceptions import FunctionNotFound, VariableNotFound
 from httprunner.loader import load_project_meta
 
 
 class TestParserBasic(unittest.TestCase):
+
+    def test_build_url(self):
+        url = parser.build_url("https://postman-echo.com", "/get")
+        self.assertEqual(url, "https://postman-echo.com/get")
+        url = parser.build_url("https://postman-echo.com", "get")
+        self.assertEqual(url, "https://postman-echo.com/get")
+        url = parser.build_url("https://postman-echo.com/", "/get")
+        self.assertEqual(url, "https://postman-echo.com/get")
+
+        url = parser.build_url("https://postman-echo.com/abc/", "/get?a=1&b=2")
+        self.assertEqual(url, "https://postman-echo.com/abc/get?a=1&b=2")
+        url = parser.build_url("https://postman-echo.com/abc/", "get?a=1&b=2")
+        self.assertEqual(url, "https://postman-echo.com/abc/get?a=1&b=2")
+
+        # omit query string in base url
+        url = parser.build_url("https://postman-echo.com/abc?x=6&y=9", "/get?a=1&b=2")
+        self.assertEqual(url, "https://postman-echo.com/abc/get?a=1&b=2")
+
+        url = parser.build_url("", "https://postman-echo.com/get")
+        self.assertEqual(url, "https://postman-echo.com/get")
+
+        # notice: step request url > config base url
+        url = parser.build_url("https://postman-echo.com", "https://httpbin.org/get")
+        self.assertEqual(url, "https://httpbin.org/get")
+
     def test_parse_variables_mapping(self):
         variables = {"varA": "$varB", "varB": "$varC", "varC": "123", "a": 1, "b": 2}
         parsed_variables = parser.parse_variables_mapping(variables)