back round 0 view raw

knotserver/git: refactor language breakdown logic #313

merged

opened by

oppi.li 5 months ago targeting master from push-snktzuwttuvu

Signed-off-by: oppiliappan me@oppi.li

options

unified split

Changed files

+68 -51

knotserver

git

language.go

routes.go

+66

knotserver/git/language.go

···

       1
       1
       +
       package git

     

       2
       2
       +
       

     

       3
       3
       +
       import (

     

       4
       4
       +
       	"context"

     

       5
       5
       +
       	"path"

     

       6
       6
       +
       

     

       7
       7
       +
       	"github.com/go-enry/go-enry/v2"

     

       8
       8
       +
       	"github.com/go-git/go-git/v5/plumbing/object"

     

       9
       9
       +
       )

     

       10
       10
       +
       

     

       11
       11
       +
       type LangBreakdown map[string]int64

     

       12
       12
       +
       

     

       13
       13
       +
       func (g *GitRepo) AnalyzeLanguages(ctx context.Context) (LangBreakdown, error) {

     

       14
       14
       +
       	sizes := make(map[string]int64)

     

       15
       15
       +
       	err := g.Walk(ctx, "", func(node object.TreeEntry, parent *object.Tree, root string) error {

     

       16
       16
       +
       		filepath := path.Join(root, node.Name)

     

       17
       17
       +
       

     

       18
       18
       +
       		content, err := g.FileContentN(filepath, 16*1024) // 16KB

     

       19
       19
       +
       		if err != nil {

     

       20
       20
       +
       			return nil

     

       21
       21
       +
       		}

     

       22
       22
       +
       

     

       23
       23
       +
       		if enry.IsGenerated(filepath, content) {

     

       24
       24
       +
       			return nil

     

       25
       25
       +
       		}

     

       26
       26
       +
       

     

       27
       27
       +
       		language := analyzeLanguage(node, content)

     

       28
       28
       +
       		if group := enry.GetLanguageGroup(language); group != "" {

     

       29
       29
       +
       			language = group

     

       30
       30
       +
       		}

     

       31
       31
       +
       

     

       32
       32
       +
       		langType := enry.GetLanguageType(language)

     

       33
       33
       +
       		if langType != enry.Programming && langType != enry.Markup && langType != enry.Unknown {

     

       34
       34
       +
       			return nil

     

       35
       35
       +
       		}

     

       36
       36
       +
       

     

       37
       37
       +
       		sz, _ := parent.Size(node.Name)

     

       38
       38
       +
       		sizes[language] += sz

     

       39
       39
       +
       

     

       40
       40
       +
       		return nil

     

       41
       41
       +
       	})

     

       42
       42
       +
       

     

       43
       43
       +
       	if err != nil {

     

       44
       44
       +
       		return nil, err

     

       45
       45
       +
       	}

     

       46
       46
       +
       

     

       47
       47
       +
       	return sizes, nil

     

       48
       48
       +
       }

     

       49
       49
       +
       

     

       50
       50
       +
       func analyzeLanguage(node object.TreeEntry, content []byte) string {

     

       51
       51
       +
       	language, ok := enry.GetLanguageByExtension(node.Name)

     

       52
       52
       +
       	if ok {

     

       53
       53
       +
       		return language

     

       54
       54
       +
       	}

     

       55
       55
       +
       

     

       56
       56
       +
       	language, ok = enry.GetLanguageByFilename(node.Name)

     

       57
       57
       +
       	if ok {

     

       58
       58
       +
       		return language

     

       59
       59
       +
       	}

     

       60
       60
       +
       

     

       61
       61
       +
       	if len(content) == 0 {

     

       62
       62
       +
       		return enry.OtherLanguage

     

       63
       63
       +
       	}

     

       64
       64
       +
       

     

       65
       65
       +
       	return enry.GetLanguage(node.Name, content)

     

       66
       66
       +
       }

+2 -51

knotserver/routes.go

···

       13
       13
        
       	"net/http"

     

       14
       14
        
       	"net/url"

     

       15
       15
        
       	"os"

     

       16
       16
       -
       	"path"

     

       17
       16
        
       	"path/filepath"

     

       18
       17
        
       	"strconv"

     

       19
       18
        
       	"strings"

     
···

       23
       22
        
       	securejoin "github.com/cyphar/filepath-securejoin"

     

       24
       23
        
       	"github.com/gliderlabs/ssh"

     

       25
       24
        
       	"github.com/go-chi/chi/v5"

     

       26
       26
       -
       	"github.com/go-enry/go-enry/v2"

     

       27
       25
        
       	gogit "github.com/go-git/go-git/v5"

     

       28
       26
        
       	"github.com/go-git/go-git/v5/plumbing"

     

       29
       27
        
       	"github.com/go-git/go-git/v5/plumbing/object"

     
···

       777
       775
        
       		return

     

       778
       776
        
       	}

     

       779
       777
        
       

     

       780
       780
       -
       	sizes := make(map[string]int64)

     

       781
       781
       -
       

     

       782
       778
        
       	ctx, cancel := context.WithTimeout(r.Context(), 1*time.Second)

     

       783
       779
        
       	defer cancel()

     

       784
       780
        
       

     

       785
       785
       -
       	err = gr.Walk(ctx, "", func(node object.TreeEntry, parent *object.Tree, root string) error {

     

       786
       786
       -
       		filepath := path.Join(root, node.Name)

     

       787
       787
       -
       

     

       788
       788
       -
       		content, err := gr.FileContentN(filepath, 16*1024) // 16KB

     

       789
       789
       -
       		if err != nil {

     

       790
       790
       -
       			return nil

     

       791
       791
       -
       		}

     

       792
       792
       -
       

     

       793
       793
       -
       		if enry.IsGenerated(filepath, content) {

     

       794
       794
       -
       			return nil

     

       795
       795
       -
       		}

     

       796
       796
       -
       

     

       797
       797
       -
       		language := analyzeLanguage(node, content)

     

       798
       798
       -
       		if group := enry.GetLanguageGroup(language); group != "" {

     

       799
       799
       -
       			language = group

     

       800
       800
       -
       		}

     

       801
       801
       -
       

     

       802
       802
       -
       		langType := enry.GetLanguageType(language)

     

       803
       803
       -
       		if langType != enry.Programming && langType != enry.Markup && langType != enry.Unknown {

     

       804
       804
       -
       			return nil

     

       805
       805
       -
       		}

     

       806
       806
       -
       

     

       807
       807
       -
       		sz, _ := parent.Size(node.Name)

     

       808
       808
       -
       		sizes[language] += sz

     

       809
       809
       -
       

     

       810
       810
       -
       		return nil

     

       811
       811
       -
       	})

     

       781
       781
       +
       	sizes, err := gr.AnalyzeLanguages(ctx)

     

       812
       782
        
       	if err != nil {

     

       813
       813
       -
       		l.Error("failed to recurse file tree", "error", err.Error())

     

       783
       783
       +
       		l.Error("failed to analyze languages", "error", err.Error())

     

       814
       784
        
       		writeError(w, err.Error(), http.StatusNoContent)

     

       815
       785
        
       		return

     

       816
       786
        
       	}

     
···

       818
       788
        
       	resp := types.RepoLanguageResponse{Languages: sizes}

     

       819
       789
        
       

     

       820
       790
        
       	writeJSON(w, resp)

     

       821
       821
       -
       	return

     

       822
       822
       -
       }

     

       823
       823
       -
       

     

       824
       824
       -
       func analyzeLanguage(node object.TreeEntry, content []byte) string {

     

       825
       825
       -
       	language, ok := enry.GetLanguageByExtension(node.Name)

     

       826
       826
       -
       	if ok {

     

       827
       827
       -
       		return language

     

       828
       828
       -
       	}

     

       829
       829
       -
       

     

       830
       830
       -
       	language, ok = enry.GetLanguageByFilename(node.Name)

     

       831
       831
       -
       	if ok {

     

       832
       832
       -
       		return language

     

       833
       833
       -
       	}

     

       834
       834
       -
       

     

       835
       835
       -
       	if len(content) == 0 {

     

       836
       836
       -
       		return enry.OtherLanguage

     

       837
       837
       -
       	}

     

       838
       838
       -
       

     

       839
       839
       -
       	return enry.GetLanguage(node.Name, content)

     

       840
       791
        
       }

     

       841
       792
        
       

     

       842
       793
        
       func (h *Handle) RepoForkSync(w http.ResponseWriter, r *http.Request) {