Urban roads are recognized to facilitate daily human activities while simultaneously being observed to shape behavioral patterns. Currently, most studies simply analyze the relationship between roads and crime distribution as part of the built environment, and few studies have thoroughly explored the impact of different structural attributes of roads on crime. In order to further explore this impact mechanism and provide guidance for the optimization of police resource allocation by front-line police departments. The operational mechanisms were investigated by which geometric and topological road attributes influence theft distribution across varied transportation modalities. Using crime data from YC District, JB City, spatial crime hotspots and high-risk road segments were initially identified through geospatial analysis. Next, the road structure was systematically decomposed into two distinct dimensions: geometric attributes and topological properties. A space syntax segment model was created to quantitatively assess geometric characteristics and traffic modality-specific topological configurations. Finally, Statistical relationships were investigated using zero-inflated negative binomial regression complemented by multiple linear regression modeling. Research has found that theft incidence is demonstrated to be positively associated with segment length, angular curvature metrics, and branch road density within the network. Elevated crime probabilities are observed in extended roadway segments exhibiting complex geometric configurations and regions characterized by dense branch road networks. The closeness and betweenness of the road's topological structure exert varying significant effects on theft crime across different traffic modes. Specifically, in pedestrian traffic modes, a significant negative correlation is observed between road closeness and betweenness and the occurrence of theft crime. In bicycle traffic modes, road closeness and betweenness are found to have a positive effect on theft crime. In electric vehicle and motor vehicle traffic modes, road closeness have a positive effect on theft crime, while in electric vehicle traffic mode, betweenness was found to have a negative effect on theft crime. These findings contribute novel insights to environmental criminology theory while offering empirically grounded recommendations for strategic police deployment and urban security management.