Text classification is major pillar of natural language processing (NLP) and plays important for the applications involving sentiment classification, information extraction, topic labelling, etc. Natural language processing has witnessed exponential growth with the advent of deep learning and transformer-based models. However, resource constrained languages like Indian languages are not able to get the benefit from these advancements due to challenges such as linguistic diversity, script heterogeneity, and scarcity of annotated data. The survey provides a comprehensive assessment of conventional and modern text classification approaches, emphasizing the adaptation of state-of-the-art deep learning techniques and transformers like mbert, IndicBERT and MuRIL to Indian languages, domain adaptation, cross-lingual techniques, and emerging multimodal strategies, highlighting key datasets. The primary focus of this review is to identify open challenges and key gaps including data augmentation, ethical considerations, and the need for inclusive datasets and to provide comprehensive review of available state of art techniques present till date.