1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
|
// ============================================================================
// Azure Content Understanding SDK Test Resources
// ============================================================================
// This Bicep template creates the following Azure resources for testing:
//
// Resources Created:
// 1. Primary Microsoft Foundry resource (Microsoft.CognitiveServices/accounts)
// - Primary resource for testing Content Understanding functionality
// - Kind: AIServices, SKU: S0
// 2. Secondary Microsoft Foundry resource (Microsoft.CognitiveServices/accounts)
// - Used as target for cross-resource copying operations (e.g., Sample15)
// - Kind: AIServices, SKU: S0
// 3. Role assignments (Microsoft.Authorization/roleAssignments)
// - Grants test application/service principal "Cognitive Services User" role
// - One for primary resource, one for secondary resource
// 4. Model deployments (Microsoft.CognitiveServices/accounts/deployments)
// - Deployed via test-resources-post.ps1 script after resource creation
// - Creates deployments for: gpt-4.1, gpt-4.1-mini, text-embedding-3-large
//
// Environment Variables Generated (outputs):
// Primary Resource:
// - CONTENTUNDERSTANDING_ENDPOINT: Primary Foundry API endpoint
//
// Primary Resource (used as source for cross-resource copy):
// - CONTENTUNDERSTANDING_SOURCE_RESOURCE_ID: Primary Foundry resource ID
// - CONTENTUNDERSTANDING_SOURCE_REGION: Primary resource region
//
// Secondary Resource (used as target for cross-resource copy):
// - CONTENTUNDERSTANDING_TARGET_ENDPOINT: Secondary Foundry API endpoint
// - CONTENTUNDERSTANDING_TARGET_RESOURCE_ID: Secondary Foundry resource ID
// - CONTENTUNDERSTANDING_TARGET_REGION: Secondary resource region
//
// Model Deployment Names:
// - GPT_4_1_DEPLOYMENT: Deployment name for gpt-4.1 model
// - GPT_4_1_MINI_DEPLOYMENT: Deployment name for gpt-4.1-mini model
// - TEXT_EMBEDDING_3_LARGE_DEPLOYMENT: Deployment name for text-embedding-3-large model
//
// Authentication:
// - Uses DefaultAzureCredential (no API keys needed)
// - Role assignments grant access via "Cognitive Services User" role
// ============================================================================
@description('The client OID to grant access to test resources.')
param testApplicationOid string
@minLength(6)
@maxLength(50)
@description('The base resource name.')
param baseName string = resourceGroup().name
@description('The location of the resource. By default, this is the same as the resource group.')
param location string = resourceGroup().location
// Role definition ID for "Cognitive Services User" role
var cognitiveServicesUserRoleId = subscriptionResourceId('Microsoft.Authorization/roleDefinitions', 'a97b65f3-24c7-4388-baec-2e87135dc908')
// Resource names
var testFoundryName = '${baseName}-foundry'
var targetFoundryName = '${baseName}-copy-target'
// Source Microsoft Foundry resource (primary resource for most tests)
resource sourceFoundry 'Microsoft.CognitiveServices/accounts@2025-04-01-preview' = {
name: testFoundryName
location: location
kind: 'AIServices'
sku: {
name: 'S0'
}
properties: {
customSubDomainName: toLower(testFoundryName)
publicNetworkAccess: 'Enabled'
}
}
// Role assignment for source resource - grants test application access
// Note: principalType is omitted to allow Azure to infer it automatically (works for both User and ServicePrincipal)
resource sourceRoleAssignment 'Microsoft.Authorization/roleAssignments@2022-04-01' = {
name: guid(resourceGroup().id, sourceFoundry.id, cognitiveServicesUserRoleId)
scope: sourceFoundry
properties: {
roleDefinitionId: cognitiveServicesUserRoleId
principalId: testApplicationOid
}
}
// Target Microsoft Foundry resource (for cross-resource copy tests, e.g., Sample15)
resource targetFoundry 'Microsoft.CognitiveServices/accounts@2025-04-01-preview' = {
name: targetFoundryName
location: location
kind: 'AIServices'
sku: {
name: 'S0'
}
properties: {
customSubDomainName: toLower(targetFoundryName)
publicNetworkAccess: 'Enabled'
}
}
// Role assignment for target resource - grants test application access
// Note: principalType is omitted to allow Azure to infer it automatically (works for both User and ServicePrincipal)
resource targetRoleAssignment 'Microsoft.Authorization/roleAssignments@2022-04-01' = {
name: guid(resourceGroup().id, targetFoundry.id, cognitiveServicesUserRoleId)
scope: targetFoundry
properties: {
roleDefinitionId: cognitiveServicesUserRoleId
principalId: testApplicationOid
}
}
// Model deployments are handled by test-resources-post.ps1 script after resource creation.
// This allows models to be deployed even if they're not available during initial Bicep deployment.
// Deployments can take 5-15 minutes to complete.
// Outputs - these become environment variables for tests
// Variable names match what ContentUnderstandingClientTestEnvironment expects
// Note: We use DefaultAzureCredential for authentication, so API keys are not needed
// Role assignments grant the test application/service principal access via the 'Cognitive Services User' role
// Construct endpoint from variable (endpoint format: https://{customSubDomainName}.services.ai.azure.com/)
// Using toLower(testFoundryName) which matches the customSubDomainName set in the resource
output CONTENTUNDERSTANDING_ENDPOINT string = 'https://${toLower(testFoundryName)}.services.ai.azure.com/'
// Primary resource outputs (used as source for cross-resource copy)
output CONTENTUNDERSTANDING_SOURCE_RESOURCE_ID string = sourceFoundry.id
output CONTENTUNDERSTANDING_SOURCE_REGION string = location
// Target resource outputs (for cross-resource copy)
// Construct endpoint from variable (endpoint format: https://{customSubDomainName}.services.ai.azure.com/)
// Using toLower(targetFoundryName) which matches the customSubDomainName set in the resource
output CONTENTUNDERSTANDING_TARGET_ENDPOINT string = 'https://${toLower(targetFoundryName)}.services.ai.azure.com/'
output CONTENTUNDERSTANDING_TARGET_RESOURCE_ID string = targetFoundry.id
output CONTENTUNDERSTANDING_TARGET_REGION string = location
// Model deployment outputs - deployment names for tests
// These match what ContentUnderstandingClientTestEnvironment expects
output GPT_4_1_DEPLOYMENT string = 'gpt-4.1'
output GPT_4_1_MINI_DEPLOYMENT string = 'gpt-4.1-mini'
output TEXT_EMBEDDING_3_LARGE_DEPLOYMENT string = 'text-embedding-3-large'
|
