1+ using Azure ;
2+ using Azure . AI . OpenAI ;
3+ using Microsoft . Azure . Functions . Worker ;
4+ using Microsoft . Extensions . Logging ;
5+ using Newtonsoft . Json ;
6+ using Newtonsoft . Json . Linq ;
7+ using Newtonsoft . Json . Schema ;
8+ using Newtonsoft . Json . Schema . Generation ;
9+ using OpenAI . Embeddings ;
10+
11+ namespace azure_project_generator
12+ {
13+ public class ProcessFile
14+ {
15+ private readonly ILogger < ProcessFile > _logger ;
16+ private readonly EmbeddingClient _embeddingClient ;
17+
18+ public ProcessFile ( ILogger < ProcessFile > logger ,
19+ EmbeddingClient embeddingClient )
20+ {
21+ _logger = logger ?? throw new ArgumentNullException ( nameof ( logger ) ) ;
22+ _embeddingClient = embeddingClient ?? throw new ArgumentNullException ( nameof ( embeddingClient ) ) ;
23+ }
24+
25+ [ Function ( nameof ( ProcessFile ) ) ]
26+ public async Task < MultipleOutput > Run (
27+ [ BlobTrigger ( "certdata/{name}" , Connection = "AzureWebJobsStorage" ) ] string content ,
28+ string name )
29+ {
30+ _logger . LogInformation ( $ "Processing blob: { name } ") ;
31+
32+ if ( string . IsNullOrWhiteSpace ( content ) )
33+ {
34+ _logger . LogError ( "Blob content is empty or whitespace." ) ;
35+ return new MultipleOutput { CertServiceDocument = null , ArchivedContent = null } ;
36+ }
37+
38+ if ( ! ValidateJsonContent ( content ) )
39+ {
40+ return new MultipleOutput { CertServiceDocument = null , ArchivedContent = null } ;
41+ }
42+
43+ var mappedServiceData = JsonConvert . DeserializeObject < MappedService > ( content ) ;
44+ if ( mappedServiceData == null )
45+ {
46+ _logger . LogError ( "Failed to deserialize content to MappedService." ) ;
47+ return new MultipleOutput { CertServiceDocument = null , ArchivedContent = null } ;
48+ }
49+
50+ string contextSentence = GenerateContextSentence ( mappedServiceData ) ;
51+ float [ ] contentVector = await GenerateEmbeddingsAsync ( contextSentence ) ;
52+
53+ var certServiceDocument = CreateCertServiceDocument ( mappedServiceData , contextSentence , contentVector ) ;
54+
55+ _logger . LogInformation ( "Document created successfully." ) ;
56+ _logger . LogInformation ( $ "Archiving blob: { name } ") ;
57+
58+ return new MultipleOutput
59+ {
60+ CertServiceDocument = certServiceDocument ,
61+ ArchivedContent = content
62+
63+ } ;
64+ }
65+
66+
67+ private async Task < string > ReadBlobContentAsync ( Stream stream )
68+ {
69+ try
70+ {
71+ using var reader = new StreamReader ( stream ) ;
72+ return await reader . ReadToEndAsync ( ) ;
73+ }
74+ catch ( IOException ex )
75+ {
76+ _logger . LogError ( ex , "Error reading blob content" ) ;
77+ return null ;
78+ }
79+ }
80+
81+ private bool ValidateJsonContent ( string content )
82+ {
83+ try
84+ {
85+ var generator = new JSchemaGenerator ( ) ;
86+ JSchema schema = generator . Generate ( typeof ( MappedService ) ) ;
87+
88+ JToken jsonContent = JToken . Parse ( content ) ;
89+ bool isValid = jsonContent . IsValid ( schema , out IList < string > messages ) ;
90+
91+ if ( ! isValid )
92+ {
93+ foreach ( var message in messages )
94+ {
95+ _logger . LogError ( $ "Schema validation error: { message } ") ;
96+ }
97+ }
98+ else
99+ {
100+ _logger . LogInformation ( "JSON content is valid against the schema." ) ;
101+ }
102+
103+ return isValid ;
104+ }
105+ catch ( JsonException ex )
106+ {
107+ _logger . LogError ( ex , "JSON parsing error during validation" ) ;
108+ return false ;
109+ }
110+ catch ( Exception ex )
111+ {
112+ _logger . LogError ( ex , "Unexpected error during JSON validation" ) ;
113+ return false ;
114+ }
115+ }
116+
117+ private string GenerateContextSentence ( MappedService data ) =>
118+ $ "The { data . CertificationCode } { data . CertificationName } certification includes the skill of { data . SkillName } . Within this skill, there is a focus on the topic of { data . TopicName } , particularly through the use of the service { data . ServiceName } .";
119+
120+ private async Task < float [ ] > GenerateEmbeddingsAsync ( string content )
121+ {
122+ try
123+ {
124+ _logger . LogInformation ( "Generating embedding..." ) ;
125+ var embeddingResult = await _embeddingClient . GenerateEmbeddingAsync ( content ) . ConfigureAwait ( false ) ;
126+ _logger . LogInformation ( "Embedding created successfully." ) ;
127+ return embeddingResult . Value . Vector . ToArray ( ) ;
128+
129+ }
130+ catch ( RequestFailedException ex )
131+ {
132+ _logger . LogError ( ex , "Azure OpenAI API request failed" ) ;
133+ throw ;
134+ }
135+ catch ( Exception ex )
136+ {
137+ _logger . LogError ( ex , "Error generating embedding" ) ;
138+ throw ;
139+ }
140+ }
141+
142+ private CertServiceDocument CreateCertServiceDocument ( MappedService data , string contextSentence , float [ ] contentVector ) =>
143+ new CertServiceDocument
144+ {
145+ id = Guid . NewGuid ( ) . ToString ( ) ,
146+ CertificationServiceKey = $ "{ data . CertificationCode } -{ data . ServiceName } ",
147+ CertificationCode = data . CertificationCode ,
148+ CertificationName = data . CertificationName ,
149+ SkillName = data . SkillName ,
150+ TopicName = data . TopicName ,
151+ ServiceName = data . ServiceName ,
152+ ContextSentence = contextSentence ,
153+ ContextVector = contentVector
154+ } ;
155+ }
156+ }
0 commit comments